Rga, a RofA-Like Regulator, Is the Major TranscriptionalActivator of the PI-2a Pilus in Streptococcus agalactiae

Shaynoor Dramsi,1,2 Sarah Dubrac,1,2 Yoan Konto-Ghiorghi,1,2,* Violette Da Cunha,1,2 Elisabeth Couve,1,2

Philippe Glaser,1,2 Elise Caliot,1,2 Michel Debarbouille,1,2 Samuel Bellais,3–5

Patrick Trieu-Cuot,1,2 and Michel-Yves Mistou1,2,6

Rapid adaptation to changing environments is key in determining the outcome of infections caused by theopportunistic human pathogen Streptococcus agalactiae. We previously demonstrated that the RofA-like protein(RALP) regulators RogB and Rga activate their downstream divergently transcribed genes, that is, the pilusoperon PI-2a and the serine-rich repeat encoding gene srr1, respectively. Characterization of the Rga regulon bymicroarray revealed that the PI-2a pilus was strongly controlled by Rga, a result confirmed at the protein level.Complementation experiments showed that the expression of Rga, but not RogB, in the double DrogB/Drgamutant, or in the clinical strain 2603V/R displaying frameshift mutations in rogB and rga genes, is sufficient torestore wild-type expression levels of PI-2a pilus and Srr1. Biofilm formation was impaired in the Drga and Drga/rogB mutants and restored on complementation with rga. Paradoxically, adherence to intestinal epithelial cellswas unchanged in the Drga mutant. Finally, the existence of several clinical isolates mutated in rga highlights theconcept of strain-specific regulatory networks.

Introduction

Streptococcus agalactiae (also known as group Bstreptococcus [GBS]) is a Gram-positive encapsulated

commensal bacterium of the human intestine that is alsopresent in the vagina of 15–30% of healthy women.2

However, in certain circumstances, GBS can become a life-threatening pathogen in neonates.9 Two distinct GBS-associated clinical syndromes, referred to as early-onsetdisease (EOD) and late-onset disease (LOD), have been rec-ognized in neonates depending on the age of the infant at thetime of disease manifestations. For EOD, epidemiologicalstudies have documented how commonly GBS are trans-mitted from ‘‘carrier’’ mothers to newborn infants.38 Trans-mission occurs by the inhalation of GBS-contaminatedamniotic or vaginal fluid during delivery, followed by bac-terial translocation across the respiratory epithelium thatrapidly progresses into subsequent systemic infection. Thelung is most likely the portal of entry. In contrast, for LOD,

the mode of transmission and the infection route haveremained elusive. LOD is characterized by bloodstreaminfection with a high risk of progression to meningitis andis strongly associated with GBS strains belonging to the‘‘hypervirulent’’ clonal complex 17.16,17,29 A recent studystrongly supports an intestinal portal of entry for LOD.39

GBS is also an emerging pathogen of adults with the elderly,immunocompromised, and those with diabetes and malig-nancies being particularly at risk.8,10 Clinical manifesta-tions are varied and include skin, soft tissue and urinarytract infections, bacteremia, pneumonia, arthritis, and en-docarditis.8,10

During the last decade, the key virulence factors of GBSinvolved in the colonization, survival, spread, and persis-tence of the bacterium within the human host have beenidentified (for a recent review, see Ref.32). In addition, theavailability of eight genome sequences accelerated our un-derstanding of the pathogen’s strategy.40 The rapid adapta-tion to the changing environments encountered during

1Institut Pasteur, Unite de Biologie des Bacteries Pathogenes a Gram-positif, Paris, France.2CNRS ERL 3526, Paris, France.3Institut Cochin, Universite Paris Descartes Faculte de Medecine, Paris, France.4CNRS UMR 8104, Paris, France.5INSERM U1016, Paris, France.6INRA, UMR 1319 Micalis, Jouy en Josas, France.*Present address: Program in Epithelial Biology and Department of Dermatology, Stanford University School of Medicine, Stanford,

California.

MICROBIAL DRUG RESISTANCEVolume 18, Number 3, 2012ª Mary Ann Liebert, Inc.DOI: 10.1089/mdr.2012.0005

286

infection is essential for the GBS survival that is needed tocolonize various tissues, obtain nutrients, or evade the hostimmune response. The genome sequences of GBS strainshave revealed the presence of 17–20 two-component systems(TCS) that can respond to changes in the external environ-ment. One of the best-characterized TCS is the CovS/CovRsystem that plays a major role in virulence.15,21 The regula-tory activity of CovR is modulated not only by its cognatesensor CovS but also by a serine/threonine kinase Stk1that inactivates CovR by phosphorylation on threonineresidue 65.22

GBS NEM316 encodes 6 ‘‘stand-alone’’ transcriptionalregulators,11 among which RogB, Rga, and Gbs1426 belongto the RofA-like protein (RALP) family (for a review, seeRef.19). RogB and Rga are the closest relatives of this familyexhibiting 47% identity, whereas Gbs1426 is more distantlyrelated, displaying 23% and 24% identity to RogB and Rga,respectively. RofA, the founding member of this family, wasfirst identified in Group A Streptococci (GAS) as a positivetranscriptional activator of the adjacent prtF gene encodingthe fibronectin binding protein SfbI.30 Nra, the second RALPregulator identified in GAS, was found to be a negativeregulator for an adjacent gene that encodes the collagen-binding protein Cpa, as well as an unlinked gene encodingthe fibronectin-binding protein PrtF2.30 RALP proteins ex-hibit strong similarities with another global regulator of Mprotein synthesis in GAS called Mga (Hondorp and McIver).Mga and RALP proteins are about 500 amino acids long andcontain 3 conserved Pfam domains: (i) HTH_Mga (PF08280);(ii) Mga (PF05043); and (iii) PRD (PF08270). Both Helix-Turn-Helix found in PF08280 and PF05043 were involved in DNAbinding27 (see Supplementary Fig. S1; Supplementary Dataare available online at www.liebertonline.com/mdr).

In the present work, we demonstrate the central role ofRga in controlling the synthesis of two important GBSNEM316 surface adhesins, namely PI-2a pilus and Srr1.Surprisingly, the adherence of an rga knock-out strain toepithelial cells was similar to that of the wild-type (WT)strain, suggesting that the display of surface proteins is ahighly dynamic process which is dependent on the expres-sion of multiple adhesins.

Materials and Methods

Bacterial strains, media, and growth conditions

The bacterial strains and plasmids used in this study arelisted in Supplementary Table S1. S. agalactiae NEM316 is aST-23 serotype III strain whose genome has been se-quenced.11 The NEM316Drga and NEM316DrogB mutantswere previously constructed.6,28 The Escherichia coli DH5a(Invitrogen) used for the cloning experiments was grown inLuria-Bertani (LB) medium. Unless otherwise specified,S. agalactiae was cultured in Todd-Hewitt (TH) broth (stand-ing filled flasks) or agar (Difco Laboratories). Erythromycinwas used at 150 mg/ml for E. coli and at 10mg/ml forS. agalactiae. Lactococcus lactis strain NZ900020 was grown inan M17 medium supplemented with 1% glucose (GM17).Heterologous expression of PI-2a pilus in the L. lactis strainwas realized as follows: The five genes gbs1478-1474 wereamplified from NEM316 genomic DNA and cloned in thelactococcal vector pOri23, a high copy number erythromycinresistance plasmid expressing the cloned gene from the

strong constitutive promoter P23.31 The primers used for theamplification of gbs1478 ( pilA), gbs1477 ( pilB), gbs1474 ( pilC),and the entire operon PI-2a (gbs1478-1474) are shown inSupplementary Table S1. The various PCR products werethen cut by the appropriate enzymes (New England Biolabs),purified, and cloned into pOri23. The resulting plasmidswere verified by DNA sequence and introduced into elec-trocompetent L. lactis NZ9000.

General DNA techniques

Standard recombinant techniques were used for nucleicacid cloning and restriction analysis.36 Plasmid DNA fromE. coli was prepared by rapid alkaline lysis using the QIA-prep Spin Miniprep Kit (Qiagen). Genomic DNA fromS. agalactiae was prepared using the DNeasy Blood and Tis-sue Kit (Qiagen). PCR was carried out with Phusion Taqpolymerase as described by the manufacturer (Finnzymes).

Construction of mutants and complemented strains

The double-mutant DrogB/Drga was constructed as de-scribed for NEM316DrogB but in the NEM316Drga geneticbackground.

The primer pairs RgaF-RgaR and RogBF-RogBR wereused to amplify the sequence encoding the rga and rogBgenes, respectively (Supplementary Table S1). The resultingPCR fragments were digested with BamH1-Pst1 and clonedinto the shuttle vector pTCVErm-Ptet to generate pRga andpRogB, respectively. The resulting plasmids were introducedby electroporation into the various strains.

Expression and purification of recombinant Sag1407

The Sag1407 DNA coding region was amplified by PCRusing genomic DNA of GBS 2603V/R as a template and primersSag1407Nco and Sag1407Bam (Supplementary Table S1). ThePCR product was digested with NcoI and BamHI enzymes andcloned into pET2817 (Tazi et al.). The resulting plasmid wasintroduced into E. coli BL21lDE3/pDIA17 for protein expres-sion. Recombinant 6xHis-SAG1407 protein was purified undernative conditions by affinity chromatography on an Ni-NTAcolumn according to the manufacturer’s recommendation(Qiagen) followed by exclusion chromatography on a Superdex200 column (GE Healthcare). Protein purity was checked onsodium dodecyl sulfate polyacrylamide gel electrophoresis(SDS-PAGE), and accurate protein concentration was deter-mined using the Bradford protein assay. Rabbit polyclonalantibodies were produced and purchased from Covalab.

Microarray analysis

We used oligo arrays carrying three oligonucleotides pergene, and this was demonstrated to be a much more sensi-tive method than the previously used DNA macro arrays.21

RNA was reverse transcribed with Superscript indirectcDNA kit (Invitrogen) and labeled with Cy5 or Cy3 (Amer-sham Biosciences) according to the supplier’s instructions.The microarray containing 6835 60-mer oligonucleotidesspecific for 2134 predicted genes of the genome of theNEM316 strain and for all intergenic regions longer than 100nucleotides has been designed using the program Oli-goArray (http://berry.engin.umich.edu/oligoarray/). Basedon these sequences a custom oligonucleotide array was

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 287

manufactured (Agilent Technologies) with a final density of15K. For hybridization, Cy3 and Cy5 target quantities werenormalized at 150 pmol.

The arrays were scanned in an Axon 4000B dual laserscanner connected to the GenePix Pro 6.0 image acquisitionand analysis software package. Slides were scanned at 635and 532 nm wavelength (respectively for Cy5 and Cy3). Aninitial prescan of the slide was performed to determine thesuccess of hybridization and to allow a definition of theboundaries of the array features. The photomultiplier (PMT)gain settings were adjusted during this step to ensure thatthe overall red-to-green ratio is similar (PMT settings are635 nm = 460, 532 nm = 425). Spots were excluded from theanalysis in the case of high local background fluorescenceslide abnormalities or weak intensity. No background wassubtracted. For the purpose of a statistical analysis, RNA wasprepared from three independent cultures, and each RNAsample was hybridized twice to the microarrays (dye swap),leading to six values. Data normalization and differentialanalysis were conducted using the R software (www.r-project.org). A global intensity-dependent normalizationusing the LOESS procedure41 was performed on a slide-by-slide basis (BioConductor package marray; www.bioconductor.org/packages/bioc/html/marray.html) to correct the dyebias. To determine the differentially expressed genes, weperformed a paired t test using the VM method (VarMixtpackage)4 together with the Benjamini and Yekutieli p-valuecorrection method.33 For each gene, 3 probes were present inthe microarray. The cut-off for the expression ratio was set toeither superior/equal to 2 or inferior/equal to 0.5, and thegeneral ratio of expression of each gene was calculated as theaverage expression ratio from the different probes. The datafor which at least two of the three probes gave a significantand noncontradictory result were taken into account.

All transcriptome data are MIAME compliant. The RAWdata have been submitted to the ArrayExpress databasemaintained at www.ebi.ac.uk/microarray-as/ae/under theAccession number: E-MEXP-3236 for the rga experiment,E-MEXP-3237 for the rogB experiment, and E-MEXP-3304for the covSR experiment. It is worth mentioning that theRga microarray analysis revealed a number of genes thatwere strongly upregulated in the Drga mutant, such asgbs0189-0190 (118- and 31-fold respectively), an observationthat was not confirmed by quantitative RT-PCR. We noticedthat they belong to a set of about 140 genes for which astrong discrepancy was repeatedly observed for the WTstrain between independant microarrays experiments and/or qRT-PCR analysis. Our hypothesis is that the expressionof these genes is unexpectedly variable from one experi-ment to another, even though the growth conditions werecautiously monitored.

Real-time quantitative PCR

Total RNAs extracted from NEM316, NEM316Drga, andNEM316DrogB were harvested at an OD600nm = 0.5 as previ-ously described.28 cDNA synthesis was performed on DNaseI treated RNA (5mg) with random hexamers (Roche) usingSuperscript II RT (Invitrogen) as recommended by themanufacturer. Real-time quantitative PCR was performed aspreviously described 28 in a 25 ml reaction volume containingcDNA, 12.5 ml iQTM SYBR Supermix (BioRad), and 1 ml gene-

specific primers (10 mM each) (Supplementary Table S2).Briefly, amplification, detection, and analysis were per-formed with the MyiQ Single-Color Real-Time iCycler PCRDetection System and the MyiQ Optical System Software(Bio-Rad). The specificity of the amplified products and theabsence of primer dimers formation were verified by gen-erating melting curves. The absence of contaminating geno-mic DNA was verified by testing each sample in controlreactions without a previous reverse-transcription step. Thecritical threshold cycle was defined for each sample. Theexpression levels of the tested genes were normalized usingthe NEM316 gyrA gene, whose transcription level did notvary under our experimental conditions. Each assay wasperformed in quadruplicate and repeated with two inde-pendent RNA samples. The change (n-fold) in the transcriptlevel was calculated as described.23

DNase I footprinting assays

DNA fragments corresponding to the promoter regions ofsrr1 (409 bp amplified with primers rga/srr1-F and -R), pilA(295 bp, rogB/pilA-F and -R) were generated by PCR usingPwo polymerase (Roche) and the indicated oligonucleotidepairs (for sequence see Supplementary Table S1). The radi-olabeling of DNA fragments was performed as previouslydescribed.5 Either purified Rga-6His or crude extracts (fromL. lactis/pAT18 and L. lactis/pAT18-Rga) binding to DNA(5 · 104 cpm per reaction) was performed at room tempera-ture in a buffer containing 25 mM NaH2PO4 pH 8, 50 mMNaCl, 2 mM MgCl2, 1 mM DTT, and 10% glycerol. DNaseItreatment and migration of the reaction mixtures were thenperformed as previously described.5

Primer extension reactions

Total RNA was used as templates for primer extensionreactions using radiolabeled specific primers of rogB(gbs1479), pilA (gbs1478), rga (gbs1529), and srr1 (gbs1530),respectively (Supplementary Table S1), as previously de-scribed.7 The corresponding Sanger DNA sequencingreactions were carried out by using the same primers andPCR-amplified fragments containing the genes upstream re-gions with the Sequenase PCR product sequencing kit (USB).

Immunoblotting analysis

Cell-wall protein extracts were prepared by harvesting50 ml of bacteria in an exponential phase (0D600 = 0.5). Thebacterial pellet was washed in ice-cold phosphate-bufferedsaline (PBS) and then in Tris-HCl buffer (50 mM, pH 7.3),and, finally, the pellet was resuspended in the mutanolysindigestion buffer (Tris-HCl 50 mM pH 7.3 supplemented with20% sucrose and a complete protease inhibitor cocktail[Roche]). Mutanolysin (Sigma) dissolved in 5,000 U ml - 1 in apotassium phosphate buffer (pH 6.2) was then added to thebacterial suspension to give a final concentration of 100 Uml - 1, and the samples were rotated for 2 hr at 37�C. Aftercentrifugation at 13,000 g for 15 min at 4�C, the supernatantscorresponding to the cell wall extracts were analyzed onSDS-PAGE or kept frozen at - 20�C. For western blotting,proteins were boiled in Laemmli sample buffer, resolved onTris-Acetate Criterion XT gradient gels 4%–12% SDS-PAGEgels (BioRad), and transferred to nitrocellulose membrane

288 DRAMSI ET AL.

(Hybond-C, Amersham). For dot-blot analysis on wholebacteria, exponentially growing bacteria were washed in PBSand resuspended in adjusted volumes of PBS to obtain similarOD600 values. The bacteria were loaded on a nitrocellulosemembrane, dried up for 20 min at room temperature, and thenblocked in PBS-milk 5% for 30 min. PilB and Srr1 were de-tected using specific rabbit polyclonal antibodies previouslyobtained6,28 at a 1:1,000 dilution and as loading control,S. agalactiae was detected using a mouse antibody raisedagainst formaldehyde-killed BM110. The secondary horse-radish peroxidase-coupled anti-rabbit secondary antibody(Zymed) was used at a 1:20,000 dilution, whereas the goatanti-mouse antibody was used at a 1:10,000 dilution. Detec-tion was performed using the Western pico chemilumines-cence kit (Thermo Scientific). Image capture and analysis wereperformed on a GeneGnome imaging system (Syngene).

Biofilm formation assays

Bacterial attachment and surface growth on polystyrenemicrotiter plates were analyzed during the growth ofS. agalactiae in an LB medium with 1% glucose supplementedwith erythromycin when necessary. Overnight culturesgrown in TH were used to inoculate the LB glucose mediumat 0.1 OD600 and after a brief vortexing, 180 ml of the cellsuspension were dispensed into 96-well plates (Costar 3799;Corning, Inc.) and incubated at 37�C for 2 or 24 hr. The OD600

of each culture was measured to ensure that they hadreached the stationary phase at similar cell densities, and thewells were washed twice in PBS using an ELx50 washer(Biotek) and air dried for 15 min. Biofilms were stained with0.1% crystal violet for 30 min (100ml per well), and the wellswere washed twice with PBS and air dried. The stainedbiomass was resuspended for quantification in ethanol/acetone (80:20), and A595 was measured. The assay wasperformed in quadriplate.

Cell culture and adherence assays

The human cell lines A549 (ATCC CCL-185) from an al-veolar epithelial carcinoma and the TC7 clone3 establishedfrom the parental colon adenocarcinoma Caco-2 were cul-tured in Quantum 286 Medium (PAA). The cells were incu-bated in 10% CO2 at 37�C and were seeded at a density of 2to 5 · 105 cells per well in 24-well tissue culture plates.Monolayers were used after 24–72 hr of incubation. Bacterialcultures from overnight cultures OD600 of 2 (*6 · 108 CFU/ml) were washed once in PBS and resuspended in DMEM.The cells were infected at a multiplicity of infection of 10bacteria per cell for 1 hr at 37�C in 10% CO2. The monolayerswere then washed four times with PBS, and the cells weredisrupted by the addition of 1 ml sterile deionized ice-coldwater and repeated pipetting. Serial dilutions of the lysatewere plated onto TH agar for the count of viable bacteria.The percent of adherence was calculated as follows: (CFU onplate count/CFU in original inoculum) · 100. Assays wereperformed in triplicate and were repeated at least thrice.

Results

Rga microarray analysis

To define the respective roles of the paralogous regula-tors Rga and RogB, we carried out a genome-wide tran-

scriptional analysis of the WT NEM316 strain and itsisogenic Drga and DrogB mutant strains. In the Drga strain,60 genes were found to be down-regulated, and 36 werefound to be up-regulated compared with the parentalNEM316 using a twofold cut-off threshold. These regulatedgenes are often grouped into operons and encode proteinsassociated to a large variety of unrelated functions, sug-gesting that Rga is a pleiotropic regulator. Among thedown-regulated genes, we identified its cognate target genesrr1, but, most importantly, the regulator-encoding generogB and its target operon gbs1478-1474 (named PI-2a inRef.35) (Table 1). In the DrogB strain, 96 genes were found tobe down–regulated, and 82 genes were found to be up-regulated. The down-regulated genes included the PI-2aoperon and the two component regulatory systems ciaRHand relPR, whereas srr1 was among the most up-regulatedgenes. RogB, similar to Rga, may have an impact on theexpression of other regulatory genes, making it difficult toidentify the genes that are directly regulated by eitherregulators from those which are indirectly regulated.

A previous macroarray analysis performed in our labo-ratory showed that rga (gbs1530), but not rogB, was twofoldinduced in the NEM316DcovSR mutant.21 This observationwas confirmed in this study by using a microarray analysis(E-MEXP-3304). We, thus, compared the Rga regulon, whichincluded 96 differentially regulated genes (twofold cut-offthreshold), with the CovSR and RogB regulons using Cluster3 and TreeView programs. The subset of 61 genes differen-tially regulated in at least two microarrays is shown in Fig. 1.The most striking group of genes likely regulated through aCovSR-dependent Rga regulation includes srr1 and the PI-2apilus operon (gbs1478-1474), the glycosyltransferase encod-ing genes gbs1605-1606, and that encoding the surface ad-hesin BibA (gbs2018) (Table 1).

A quantitative analysis of the results shows that the ex-pression of the PI-2a operon is positively regulated by bothRga and RogB, but that the impact of the rga deletion is 10times that of the rogB mutation.

Rga is the master regulator of the pilus operon

In order to determine the hierarchy between Rga andRogB in the transcriptional activation of the PI-2a pilusoperon (gbs1478-1474) and srr1 (gbs1529), quantitative re-verse transcription polymerase chain reaction (RT-PCR)experiments on srr1 and pilB were performed using the WTstrain NEM316 and its isogenic Drga and DrogB mutantstrains. The results shown in Fig. 2 demonstrated that Rgais the major transcriptional activator of PI-2a pilus and srr1transcription ( > 100-fold on pilB, the second gene of theoperon, encoding the major pilin). As previously described,RogB only slightly activates pilB transcription two- tothreefold.6,13 We also observed a slight repression in srr1transcription by RogB, confirming the transcriptomic data.

By dot-immunoblotting with specific polyclonal anti-bodies against Srr1 and PilB on exponentially growing bac-teria, we determined Srr1 and PilB protein levels in the WTstrain NEM316 and the two isogenic mutants Drga andDrogB. As a control for bacterial loading, we used an anti-body against formaldehyde-killed GBS bacteria. No signalcould be detected in the Drga mutant, indicating that Rga isthe major activator of both Srr1 and PilB synthesis (Fig. 3).

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 289

Trans-complementation with a plasmid-borne rga restoredboth Srr1 and PilB synthesis (Fig. 3).

The complex interaction between RogB and Rga promptedus to construct a double Drga/rogB mutant derivative of theNEM316 strain. We then introduced plasmid-borne rga orrogB under the control of a constitutive promoter in both theWT and double-mutant strains. Western-blot and dot-blotanalyses showed that (i) the overexpression of rogB or rga inthe WT strain increased the formation of PilB polymers; and(ii) the expression of rga only, but not rogB, in the Drga/rogBstrain was sufficient to restore WT levels of pilus synthesis(Fig. 4). Similarly, overexpression of rga in the WT strainincreased the expression level of Srr1 at the cell surface andwas sufficient to restore the WT levels of Srr1 in the Drga/rogB strain (data not shown).

The analysis of the rga sequence in the three complete GBSgenomes and four partially sequenced genomes revealedthat strains A909, 2603V/R, and CJB111 likely producedtruncated Rga, resulting from frameshift mutation and/ormutation creating an internal stop codon (SupplementaryFig. S1). Interestingly, the GBS strain 2603V/R responsiblefor adult invasive disease also exhibited a frameshift muta-tion in the rogB gene, thus constituting a natural rga/rogBdouble mutant. We first showed by immunoblotting analy-ses that 2603V/R did not synthesize any detectable levels ofSrr1 and PilB proteins (not shown). By introducing in trans afunctional copy of rga (pRga) in the 2603V/R strain, we re-stored the expression of both PilB polymers and Srr1,whereas the trans-expression of rogB did not do so (Fig. 4 forPilB, not shown for Srr1). Altogether, these results indicate

Table 1. List of the Annotated Genes Most Differentially Expressed in the Drga, DrogB,

and DCovRS Mutants Relative to the Parental Strain NEM316a

GenebFunctional

classificationc Alternative name, functionRatio

Drga/WTRatio

DrogB/WTRatio

DcovRS/WT

Genes activated by Rgagbs1529* 1.8 srr1, Ser rich repeat glycoprotein (LPXTG) 0.04 3.09 4.97d

gbs1474* 1.8 pilC, pilus-anchoring subunit (LPXTG) 0.09 0.98 3.91d

gbs1475* 1.8 srtC4, pilus-associated class C sortase 0.11 0.69 3.88d

gbs1476* 1.8 srtC3, pilus-associated class C sortase 0.09 0.69 3.57d

gbs1477* 1.8 pilB, major pilin subunit (LPXTG) 0.03e 0.4 3.81d

gbs1478* 1.8 pilA, pilus-associated adhesin (LPXTG) 0.07e 0.4 3.36d

gbs1479 3.5.2 rogB, transcriptional regulator RALP family 0.31 0.3 0.88gbs1432 2.3 Carbamoyl-phosphate syntase, LU 0.37 0.88 1.01gbs1433 2.3 Carbamoyl-phosphate syntase, SU 0.27 0.79 1.02gbs1434 3.5.2 Pyrimidine operon regulatory protein 0.30 0.75 0.99gbs1435 3.6 Ribosomal pseudouridine synthase, RluA 0,08 0.89 1.12gbs1436* 1.6 lsp, Signal peptidase II 0.06 0.78 1.22gbs1437* 3.5.2 Transcriptional regulator, LysR family 0.07 0.73 1.11gbs2083 1.2 Arginine/ornithine antiporter 0.07 0.94 0.12gbs2084* 2.2 Carbamate kinase (CK) 0.09 1 0.11gbs2085* 2.2 Ornithine carbamoyltransferase 1 0.11 0.94 0.22gbs2122 2.2 Arginine deiminase 0.44 0.61 0.23gbs2123 5.2 Acetyl transferase GNAT family 0.45 0.27 0.17gbs2124 2.2 Ornithine carbamoyltransferase 2 0.18 0.31 0.15gbs2125 1.2 Arginine/ornithine antiporter 0.21 0.35 0.34gbs2126 2.2 Carbamate kinase (ArcC) 0.26 0.35 0.57gbs1995 5.2 Phospholipid phosphatase 0.2e 0.36 0.78gbs0322 2.2 cysK, Cysteine synthase 0.3 2.06 0.88gbs1859 2.3 PurA, similar to adenylosuccinate synthase 0.33 1.14 0.39gbs1376 4.1 clpL, ATP dependant Clp protease 0.37e 0.28 0.69gbs1399 5.2 relP, RelA/SpoT domain protein 0.47e 0.33 0.79gbs1734 1.2 scrA, PTS system, sucrose-specific EIIABC 0.24 0.29 0.45gbs1330 1.2 lacE, lactose-specific PTS system, EIIBC 0.34 0.63 0.35gbs1331 1.2 Lactose-specific PTS system, EIIA 0.45 0.59 0.68gbs1444 5.2 Ser-rich repeat protein 0.35 0.67 0.46gbs2018 1.8 BibA, adhesin (LPXTG) 0.49 0.66 77.17d

gbs1605 5.2 Glycosyltransferase, family 2 0.44e 0.53 101.19d

gbs1606 5.2 Glycosytransferase, family 2 0.43e 0.48 77.63d

gbs1607 5.2 Spore coat protein CotH 0.31e 0.55 68.83d

gbs1608 6 Membrane-associated protein 0.57 0.5 72.43d

gbs1609 6 VTC-domain containing protein 0.46 0.57 76.51d

aRNAs were extracted from the bacteria collected at the mid-exponential phase of growth in Todd–Hewitt broth (OD600 = 0.3).bAsterisk (*) indicates the few chosen genes whose differential expression has been validated by quantitative reverse transcription

polymerase chain reaction, thereby confirming the microarray results.cThe functional classification is that of the SagaList server: http://genolist.pasteur.fr/SagaList/help/function-codes.html.dRga-dependent CovRS regulation.eRogB-dependent Rga regulation.RALP, RofA-like protein; GBS, group B streptococcus; WT, wild-type.

290 DRAMSI ET AL.

that the expression of rga is necessary and sufficient for bothPI-2a pilus and Srr1 expression.

Characterization of PI-2a and Srr1 promoters

To gain further insights in the molecular mechanismsunderlying the Rga-dependant regulation, transcription startsites were localized in the rogB-pilA and rga-srr1 intergenicregions (Supplementary Fig. S2). Single nonoverlapping

transcriptional start sites (TSS) were found upstream pilA(the first gene of the PI-2a operon), rogB, srr1, and rga genes,suggesting no transcriptional interference between the di-vergently transcribed genes (Supplementary Fig. S2).

In vitro binding assays were performed to test whetherthe Rga-dependant activation results from the direct in-teraction of Rga with its target promoters. Our many at-tempts to express and purify the recombinant Rga proteinin E. coli using different systems (Rga-6His or MBP-Rga or

FIG. 1. Comparison of the Rga regulon to the CovSR and RogB regulons using Cluster 3.0 and Java Tree View represen-tation. Among the 97 differentially regulated genes in the Rga regulon, a subset of 61 genes whose expression varies in theCovSR or RogB microarrays are shown. Upregulated genes are indicated in red, downregulated genes are indicated in green,and nonvariable genes are shown in black.

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 291

RgA-intein) have failed. In all cases, the protein was eitherinsoluble or degraded (data not shown). We, therefore,cloned the entire rga with a C-terminal histidyl tag in themulticopy vector pAT18 giving rise to pRga-6His. Thisplasmid, able to complement the Drga mutant (not shown),was expressed in the L. lactis strain NZ9000 under thecontrol of a constitutive promoter. The recombinant proteinRga-6His was purified from L. lactis, and its identity wasverified by mass spectrometry analysis (data not shown).DNAse I footprinting assays were carried out using eitherthe purified Rga-6His protein (20 mg) or total protein ex-tracts (25–100 mg) from L. lactis (pRga-6His) or L. lactis(pAT18). As shown in Fig. 5, two close DNA sequences inthe srr1 promoter region from - 100 to - 69 and from - 64to - 45 relative to the TSS were protected with the totalextracts of L. lactis (pRga-6His). Protection was detectedwith neither the control L. lactis (pAT18) extracts, nor thepurified protein Rga-6His. The same experiment was donewith the PI-2a promoter region, and no protection againstDNase I digestion was observed (data not shown). Theseresults show that (i) srr1 is directly regulated by Rga, whilethe PI-2a operon is probably indirectly regulated; and (ii)that purification of Rga is detrimental to its DNA-bindingactivity, either because of protein instability or due to theloss of a cytoplasmic cofactor.

FIG. 2. Expression of the genes pilB and srr1 in Strepto-coccus agalactiae DrogB (top panel) and Drga (bottom panel).The transcriptional analysis was performed by quantitativereverse transcription polymerase chain reaction and nor-malized relative to the gyrA gene. The figure indicates thecalculated ratio of the expression in the mutant strains rela-tive to that in the NEM316 WT strain. WT, wild-type.

FIG. 3. Analysis of surface exposure of Srr1 and PilB pro-teins in WT NEM316, Drga, and DrogB strains by im-munodot-blot. 2 and 6ml of whole bacterial cells suspensionin phosphate buffer saline (OD600nm = 1) harvested in theexponential phase (OD600nm = 0.5) were spotted on nitrocel-lulose. Membranes were hybridized with specific anti-Srr1and anti-PilB rabbit polyclonal antibodies. Anti-GBS antise-rum raised in mice was used as a loading control. Themembranes were next incubated with horseradish peroxi-dase-conjugated secondary antibodies and revealed usingthe Western pico chemiluminescence kit. GBS, group Bstreptococcus.

FIG. 4. Western blot analysis of cell-wall anchored pili inthree S. agalactiae strains overexpressing rogB or rga. NEM316and its isogenic Drga/rogB mutant strains (A) and S. aga-lactiae 2603V/R strain (B) harboring the empty vector or theplasmids carrying rogB or rga. All the strains were grown inthe presence of erythromycin. Mutanolysin-extracted cell-wall-associated proteins were separated on 4%–12% gradientsodium dodecyl sulfate polyacrylamide gel electrophoresisand detected after transfer on nitrocellulose with specificanti-PilB (A) of anti-Sag1407 (B) antisera. The molecularweight markers are indicated on the left (kDa).

292 DRAMSI ET AL.

Biofilm formation is abolished in the Drgamutant and restored on complementation

A major role of GBS PI-2a pilus is its implication in bio-film formation.18,34 Indeed, expression of the NEM316 PI-2apilus operon in the nonpathogenic L. lactis strain NZ9000conferred to the recombinant bacteria the ability to formbiofilm (Supplementary Fig. S3 and Ref.34). We, therefore,tested the isogenic Drga and DrogB mutants for their abilityto form biofilm on polystyrene plates using LB glucosemedium. As shown in Fig. 6A, biofilm formation is slightlyreduced in the DrogB mutant but strongly impaired in theDrga mutant as compared with the parental NEM316 strain.The complementation of DrogB and Drga by plasmid over-expression of rogB and rga, respectively, restored biofilmformation (Fig. 6B). Finally, we showed that the over-expression of rga in the Drga/rogB was sufficient to confer tothe strain the ability to form biofilm, whereas the over-expressing vector alone or rogB did not (Fig. 6C). Similarly,the overexpression of rga in GBS strain 2603V/R conferredto this GBS clinical isolate the capacity to form biofilm inour experimental conditions. Finally, the overexpression ofrga in the WT strain increased the levels of biofilm as

compared with the WT strain carrying the vector alone orthe plasmid carrying rogB (Fig. 6C).

The second important role assigned to GBS pilus PI-2A isthe promotion of adhesion to alveolar epithelial cellsA549,6,18 whereas Srr1 was shown to be a surface adhesin.28

We, thus, tested the adhesive capacity of the Drga mutant toA549 alveolar cells and TC7 intestinal cells. Surprisingly, wedid not found any significant difference for adherence tothese cell lines between the Drga mutant and the WT strain(Fig. 7 and data not shown). This led us to hypothesize thatexpression and/or display of other surface proteins in theDrga mutant compensate for the absence of GBS adhesinsSrr1 and PI-2a pilus.

Discussion

We show here that the PI-2a pilus and Srr1, two majorsurface components previously characterized in our labora-tory, are under the positive control of a common transcrip-tional regulator Rga belonging to the RALP family. Wedemonstrated the pre-eminence of Rga over RogB, the sec-ond closest RALP paralog exhibiting 47% identity with Rgaand previously shown to activate the transcription of the

FIG. 5. DNase I footprinting analysis of Rga binding to the srr1 promoter region. (A) DNase I protection assay on the srr1promoter region using crude extracts of Lactococcus lactis harboring either the pAT18 empty vector or the pAT18-rga plasmid.Lanes contain *0.5 pmol (5 · 104 cpm per reaction) of labeled template strand of srr1 ( - 176 to + 232 relative to the tran-scription start site). Lane 1: no protein; lanes 2 and 5: 25 mg; lanes 3 and 6: 50 mg; lanes 4 and 7: 100 mg of L. lactis/pAT18 crudeextract in lanes 2, 3, and 4 and L. lactis/pAT18-rga crude extract in lanes 5, 6, and 7. Lane 8: A + G Maxam and Gilbertreaction. Brackets indicate the regions protected by Rga from DNase I cleavage. (B) Nucleotide sequences of the rga promoterregion. The initiation site of transcription has been mapped (Supplementary Fig. S2) and indicated as + 1, and the - 10 and -35 regions are shaded. The regions protected from DNase I digestion on Rga binding are indicated.

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 293

pilus operon PI-2a.6,13 It is worth noting that the synthesis ofthese adhesins is partially dependent on the TCS CovSR,which acts by negatively controlling the transcription of rga.A working model of regulation is depicted in Fig. 8.

Relatively little is known about the regulation of virulencefactors in GBS besides the central role of CovSR, a TCS. Weand others previously showed that the response regulatorCovR is a pleiotropic regulator of virulence genes whichbinds to the promoter regions of the genes that it con-trols.15,21 A consensus CovR recognition sequence has beendetermined (5¢-TATTTTAAT-3¢) using both DNAse I foot-printing and computational analyses.21 Interestingly, the rgagene was twofold induced in the DcovRS strain, indicatingthat a part of the response of S. agalactiae to the covR muta-tion may be indirect. By carefully examining the rga pro-moter region, we found a putative binding site for CovRlocated 40 bp from the TSS with two mismatches in the rec-ognition sequence.

Rga and RogB belong to the Mga surperfamily which in-cludes the RALP family or RALPs originally described inGAS (for a recent review, see Ref.25). These transcriptionalregulators coined as ‘‘stand-alone’’ response regulators werepreviously shown to control global virulence regulons in

response to changing environmental conditions.14 Similar tothe other members of this family of regulators, Rga is pre-dicted to exhibit two successive helix-turn-helix motifs at itsN-terminus, which is consistent with a DNA-binding func-tion. Only two poorly defined consensus sites have beendefined for Mga and RofA.12,26 Interestingly the DNA se-quence protected by Rga found upstream from srr1 (boxes Iand II) is reminiscent of the RofA-binding site found withinthe intergenic region shared by rofA and prtF at a similarposition relative to the TSS.12 Computational analyses of thevarious promoter regions of Rga target genes were per-formed but did not reveal any clear consensus sequence,suggesting a complex regulation.

During the preparation of this article, another study re-porting the role of Rga in pilus formation in S. agalactiaeNEM316 was published.37 These authors claim that recom-binant Rga purified from E. coli can bind to the promoterregions of pilA and fbsA. We were unable to see any con-vincing gel shift in the pilA and srr1 promoter regions withpurified Rga under various conditions (not shown). Usingthe DNAse I protection assay, a protected region in the srr1promoter region was defined with total extracts of L. lactisexpressing rga compared with the L. lactis carrying the vector

FIG. 6. Role of transcriptional regulators Rga and RogB in biofilm formation. S. agalactiae NEM316 and its isogenic Drga andDrogB mutants (A), DrogB, or Drga complemented strains (B), NEM316, NEM316Drga/rogB, and 2603V/R overexpressingrogB or rga as compared with the corresponding strain harboring the empty plasmid (vec) (C). Bacteria were grown in 96-wellpolystyrene plates in Luria-Bertani supplemented with 1% glucose at 37�C for 24 hr. The adherent bacteria were stained withCV, and quantification was performed by measuring the absorbance at 595 nm. For the bacteria harboring the plasmids, theexperiments of biofilm formation were performed in the presence of erythromycin, which affects the final biomass and resultsin overall less biofilm (compare B and A). The results displayed in the histograms correspond to the average of threeindependent experiments (error bars are standard deviations). CV, crystal violet.

294 DRAMSI ET AL.

alone. However, no protection was observed with the re-combinant purified Rga-6His alone. Collectively, our resultson Rga suggest that other bacterial components are neces-sary to promote the binding of Rga to certain of its targetpromoters.

A genome-wide transcriptional analysis indicated thatRga controls not only the expression of genes encoding thesurface proteins but also additional aspects of bacterialphysiology, in particular, arginine catabolism (gbs2083-2085;gbs2122-2126). Both gbs2083-2085 and gbs2122-2126 operonsencode genes showing homologies to those found in the ar-ginine deiminase pathway (ADI). This multi-enzyme com-plex catalyzes the conversion of arginine to ornithine,ammonium, and carbon dioxide with the concomitant pro-duction of ATP. ADI-expressing bacteria were shown to bebetter not only in overcoming oxygen and nutrient starvationbut also in surviving in acidic environments.24 From themicroarray data (Table 1), it appears that Rga strongly acti-vates the ADI pathway. We, therefore, tested the survival ofGBS NEM316 and its isogenic mutant strain Drga in variouspH conditions (6.0, 5.5, 5 and 4). We found that the growth ofthe Drga mutant was strongly impaired in acidic conditions,with a 2-log reduction in viability after 2 hr at pH 4 com-pared with the WT strain (data not shown). It is, thus,tempting to speculate that Rga may play a major role duringvaginal colonization of the GBS strain NEM316 and that thepilus PI-2A and Srr1 are the major adhesins involved in thisprocess.

We and others previously showed a major role of GBS PI-2a pilus in biofilm formation.18,34 Here, we further

strengthen this result by demonstrating a strict correlationbetween the PI-2a expression and the ability of the differentmutants tested to form biofilm (Fig. 6 and SupplementaryFig. S3). We previously showed that both PI-2a pilus and thesurface glycoprotein Srr1 are bacterial adhesins for host ep-ithelial cells.6,18,28 To our surprise, the Drga mutant, which isstrongly impaired for both PI-2a and Srr1 synthesis, wasfound to be as adherent as the parental strains on humanepithelial cells A549 and TC7. We suggest a profound re-modeling of the bacterial surface in the Drga mutant with theexposition of other potential adhesins such as FbsA (Samenet al.), a major fibrinogen binding protein, that could poten-tially compensate for the loss of PI-2a and Srr1. Global sur-face proteome changes might also explain the behaviour oftwo other NEM316 derivatives, Dgbs1478-74 and DsrtC3C4mutant in the PI-2a locus, which were found to be as ad-herent as the WT strain to epithelial cells (S. Dramsi, un-published data). Overall these results suggest that theadherence of GBS to epithelial cells is probably mediated bymultiple adhesins and that the removal of one or severaladhesins from the GBS cell surface is often compensated bythe display of other surface proteins involved in the samefunction, proving the robustness of the system.

In contrast, the biofilm formation in LB glucose is strictlydependent on PI-2a pilus synthesis in GBS. Moreover, therecombinant L. lactis strains overexpressing the PI-2a pilusoperon are able to form biofilm34 (Supplementary Fig. S3).We also noticed that overexpression of rga in the Drga strainthat increased biofilm formation decreased the adher-ence level to epithelial cells, suggesting, in these conditions, afine-tuned balance between bacteria-bacteria and bacteria-host interactions (data not shown). Of note, the recombi-nant L. lactis strains overexpressing the PI-2a pilus or the

FIG. 8. Working model for Rga regulation and its integra-tion in other regulatory networks. Strong transcriptionalactivation of srr1 and PI-2a operon by Rga (black thick lines)and low activation of the PI-2a operon by RogB (gray thinlines). RogB also activates its own transcription and slightlyrepresses srr1. At an upper level, the CovS/R two-compo-nent system, central for virulence in GBS, negatively regu-lates rga transcription. STK, serine threonine kinase.

FIG. 7. Adhesion of S. agalactiae NEM316 and its isogenicDrogB, Drga, and Drga/rogB mutants to the alveolar epithelialA549 cell line. Monolayers were infected at a multiplicity ofinfection of 10 bacteria per cell for 1 hr at 37�C. The adher-ence frequencies were calculated from the number of bacteriathat remained attached to the cell after the incubation periodand three washes with regard to the number of inoculatedbacteria. The level of adherence of the WT strain is arbitrarilyreported as 100, and the level of adherence of the variousmutants are relative values. The results are presented asmean value ( – SD) from three independent experiments, andeach strain is tested in triplicate.

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 295

pilus-associated adhesin PilA were unable to adhere to epi-thelial cells as compared with the control L. lactis straincarrying the empty plasmid (Supplementary Fig. S3).

Our study was performed on the NEM316 strain, but wenoticed that, among the other available GBS genomes, atleast three strains (A909, CJB111, and 2603V/R) display aframeshift mutation in the rga gene at various positions,leading to a truncated protein. Similarly, the GBS strains2603V/R and 515 display a frameshift mutation in the rogBgene (Supplementary Fig. S1). It is worth noting that in theGBS strain COH1, which belongs to the hypervirulent clonalcomplex ST-17, the chromosomal locus PI-2a is replaced byPI-2b and that the srr1 locus is missing; consequently, thisstrain expresses neither Rga nor RogB regulators.1 Theseobservations reveal that the GBS clinical isolates exhibit adiverse repertoire of surface components and associatedtranscriptional regulators.

This work leads us to consider the bacterial surface as adynamic structure susceptible to unpredictable remodelingthat may have important functional consequences. Thus, theconcept of pathogenicity in streptococci appears more com-plex than previously thought.

Acknowledgments

This work was supported in part by grant ANR-06-PA-THO-001-01 from the Agence Nationale de la Recherche‘‘ERA-NET PathoGenoMics’’ (to P. T.-C.) and by the InstitutPasteur. The authors thank Tarek Msadek for his help andconstant support during the purification of recombinant Rgaprotein and Mathieu Brochet for the preliminary results us-ing macroarrays. They also thank Agnes Fouet for a criticalreading of the article.

Disclosure Statement

No competing financial interests exist.

References

1. Brochet, M., E. Couve, M. Zouine, T. Vallaeys, C. Rusniok,

M.C. Lamy, C. Buchrieser, P. Trieu-Cuot, F. Kunst, C. Poyart,

et al. 2006. Genomic diversity and evolution within the speciesStreptococcus agalactiae. Microbes Infect. 8:1227–1243.

2. Campbell, J.R., S.L. Hillier, M.A. Krohn, P. Ferrieri, D.F.

Zaleznik, and C.J Baker. 2000. Group B streptococcal colo-nization and serotype-specific immunity in pregnant womenat delivery. Obstet. Gynecol. 96:498–503.

3. Chantret, I., A. Rodolosse, A. Barbat, E. Dussaulx, E. Brot-

Laroche, A. Zweibaum, and M. Rousset. 1994. Differentialexpression of sucrase-isomaltase in clones isolated fromearly and late passages of the cell line Caco-2: evidence forglucose-dependent negative regulation. J. Cell. Sci. 107 (Pt

1):213–225.4. Delmar, P., S. Robin, and J.J. Daudin. 2005. VarMixt: effi-

cient variance modelling for the differential analysis ofreplicated gene expression data. Bioinformatics 21:502–508.

5. Derre, I., G. Rapoport, and T. Msadek. 1999. CtsR, a novelregulator of stress and heat shock response, controls clp andmolecular chaperone gene expression in gram-positive bac-teria. Mol. Microbiol. 31:117–131.

6. Dramsi, S., E. Caliot, I. Bonne, S. Guadagnini, M.C. Pre-

vost, M. Kojadinovic, L. Lalioui, C. Poyart, and P. Trieu-

Cuot. 2006. Assembly and role of pili in group B strepto-cocci. Mol. Microbiol. 60:1401–1413.

7. Dubrac, S., I.G. Boneca, O. Poupel, and T. Msadek. 2007.New insights into the WalK/WalR (YycG/YycF) essentialsignal transduction pathway reveal a major role in control-ling cell wall metabolism and biofilm formation in Staphy-lococcus aureus. J. Bacteriol. 189:8257–8269.

8. Edwards, M.S., and C.J. Baker. 2005. Group B streptococcalinfections in elderly adults. Clin. Infect. Dis. 41:839–847.

9. Edwards, M.S., V. Nizet, and C.J. Baker. 2006. Group Bstreptococcal infections. In: JS Remington, JO Klein, C Wilsonand CJ Baker (eds.), Infectious Diseases of the Fetus and New-born Infant. Saunders Elsevier, Philadelphia, PA, pp. 403–464.

10. Farley, M.M. 2001. Group B streptococcal disease in non-pregnant adults. Clin. Infect. Dis. 33:556–561.

11. Glaser, P., C. Rusniok, C. Buchrieser, F. Chevalier, L.

Frangeul, T. Msadek, M. Zouine, E. Couve, L. Lalioui, C.

Poyart, et al. 2002. Genome sequence of Streptococcus aga-lactiae, a pathogen causing invasive neonatal disease. Mol.Microbiol. 45:1499–1513.

12. Granok, A.B., D. Parsonage, R.P. Ross, and M.G. Caparon.

2000. The RofA binding site in Streptococcus pyogenes isutilized in multiple transcriptional pathways. J. Bacteriol.182:1529–1540.

13. Gutekunst, H., B.J. Eikmanns, and D.J. Reinscheid. 2003.Analysis of RogB-controlled virulence mechanisms and generepression in Streptococcus agalactiae. Infect. Immun. 71:5056–5064.

14. Hondorp, E.R., and K.S. McIver. 2007. The Mga virulenceregulon: infection where the grass is greener. Mol. Microbiol.66:1056–1065.

15. Jiang, S.M., M.J. Cieslewicz, D.L. Kasper, and M.R. Wes-

sels. 2005. Regulation of virulence by a two-componentsystem in group B streptococcus. J. Bacteriol. 187:1105–1113.

16. Jones, N., J.F. Bohnsack, S. Takahashi, K.A. Oliver, M.S.

Chan, F. Kunst, P. Glaser, C. Rusniok, D.W. Crook, R.M.

Harding, et al. 2003. Multilocus sequence typing system forgroup B streptococcus. J. Clin. Microbiol. 41:2530–2536.

17. Jones, N., K.A. Oliver, J. Barry, R.M. Harding, N. Bisharat,

B.G. Spratt, T. Peto, and D.W. Crook. 2006. Enhanced in-vasiveness of bovine-derived neonatal sequence type 17group B streptococcus is independent of capsular serotype.Clin. Infect. Dis. 42:915–924.

18. Konto-Ghiorghi, Y., E. Mairey, A. Mallet, G. Dumenil, E.

Caliot, P. Trieu-Cuot, and S. Dramsi. 2009. Dual role forpilus in adherence to epithelial cells and biofilm formation inStreptococcus agalactiae. PLoS Pathog. 5:e1000422.

19. Kreikemeyer, B., K.S. McIver, and A. Podbielski. 2003.Virulence factor regulation and regulatory networks inStreptococcus pyogenes and their impact on pathogen-hostinteractions. Trends Microbiol. 11:224–232.

20. Kuipers, O.P., P.G.G.A. de Ruyter, M. Kleerebezem, and

W.M. de Vos. 1998. Quorum sensing-controlled gene ex-pression in lactic acid bacteria. J. Biotechnol. 64:15–21.

21. Lamy, M.C., M. Zouine, J. Fert, M. Vergassola, E. Couve, E.

Pellegrini, Glaser P., F. Kunst, T. Msadek, P. Trieu-Cuot,

et al. 2004. CovS/CovR of group B streptococcus: a two-component global regulatory system involved in virulence.Mol. Microbiol. 54:1250–1268.

22. Lin, W.J., D. Walthers, J.E. Connelly, K. Burnside, K.A.

Jewell, L.J. Kenney, and L. Rajagopal. 2009. Threoninephosphorylation prevents promoter DNA binding of theGroup B Streptococcus response regulator CovR. Mol. Mi-crobiol. 71:1477–1495.

296 DRAMSI ET AL.

23. Livak, K.J., and T.D. Schmittgen. 2001. Analysis of relativegene expression data using real-time quantitative PCR andthe 2(-Delta Delta C(T)) method. Methods 25:402–408.

24. Marquis, R.E., G.R. Bender, D.R. Murray, and A. Wong.

1987. Arginine deiminase system and bacterial adaptation toacid environments. Appl. Environ. Microbiol. 53:198–200.

25. McIver, K.S. 2009. Stand-alone response regulators con-trolling global virulence networks in Streptococcus pyogenes.Contrib. Microbiol. 16:103–119.

26. McIver, K.S., A.S. Heath, B.D. Green, and J.R. Scott. 1995.Specific binding of the activator Mga to promoter sequencesof the emm and scpA genes in the group A streptococcus. J.Bacteriol. 177:6619–6624.

27. McIver, K.S., and R.L. Myles. 2002. Two DNA-bindingdomains of Mga are required for virulence gene activation inthe group A streptococcus. Mol. Microbiol. 43:1591–1601.

28. Mistou, M.Y., S. Dramsi, S. Brega, C. Poyart, and P. Trieu-

Cuot. 2009. Molecular dissection of the secA2 locus of groupB Streptococcus reveals that glycosylation of the Srr1 LPXTGprotein is required for full virulence. J. Bacteriol. 191:4195–4206.

29. Musser, J.M., S.J. Mattingly, R. Quentin, A. Goudeau, and

R.K. Selander. 1989. Identification of a high-virulence cloneof type III Streptococcus agalactiae (group B Streptococcus)causing invasive neonatal disease. Proc. Natl. Acad. Sci.U. S. A. 86:4731–4735.

30. Podbielski, A., M. Woischnik, B.A. Leonard, and K.H.

Schmidt. 1999. Characterization of nra, a global negativeregulator gene in group A streptococci. Mol. Microbiol.31:1051–1064.

31. Que, Y.A., J.A. Haefliger, P. Francioli, and P. Moreillon.

2000. Expression of Staphylococcus aureus clumping factor Ain Lactococcus lactis subsp. cremoris using a new shuttlevector. Infect. Immun. 68:3516–3522.

32. Rajagopal, L. 2009. Understanding the regulation of GroupB Streptococcal virulence factors. Future Microbiol. 4:201–221.

33. Reiner, A., D. Yekutieli, and Y. Benjamini. 2003. Identify-ing differentially expressed genes using false discovery ratecontrolling procedures. Bioinformatics 19:368–375.

34. Rinaudo, C.D., R. Rosini, C.L. Galeotti, F. Berti, F. Necchi,

V. Reguzzi, C. Ghezzo, J.L. Telford, G. Grandi, and D.

Maione. 2010. Specific involvement of pilus type 2a in bio-film formation in group B Streptococcus. PLoS One 5:e9216.

35. Rosini, R., C.D. Rinaudo, M. Soriani, P. Lauer, M. Mora, D.

Maione, A. Taddei, I. Santi, C. Ghezzo, C. Brettoni, et al.

2006. Identification of novel genomic islands coding forantigenic pilus-like structures in Streptococcus agalactiae. Mol.Microbiol. 61:126–141.

36. Sambrook, J., E.F. Fritsch, and T. Maniatis. 1989. MolecularCloning: A Laboratory Manual. Cold Spring Harbor La-boratory, New-York, NY.

37. Samen, U., B. Heinz, H. Boisvert, B.J. Eikmanns, D.J. Re-

inscheid, and F. Borges. 2011. Rga is a regulator of adher-ence and pili formation in Streptococcus agalactiae.Microbiology 157(Pt 8): 2319–2327.

38. Schuchat, A. 1998. Epidemiology of group B streptococcaldisease in the United States: shifting paradigms. Clin. Mi-crobiol. Rev. 11:497–513.

39. Tazi, A., O. Disson, S. Bellais, A. Bouaboud, N. Dmytruk,

S. Dramsi, M.Y. Mistou, H. Khun, C. Mechler, I. Tardieux,

et al. 2010. The surface protein HvgA mediates group Bstreptococcus hypervirulence and meningeal tropism in ne-onates. J. Exp. Med. 207:2313–2322.

40. Tettelin, H., V. Masignani, M.J. Cieslewicz, C. Donati, D.

Medini, N.L. Ward, S.V. Angiuoli, J. Crabtree, A.L. Jones,

A.S. Durkin, et al. 2005. Genome analysis of multiplepathogenic isolates of Streptococcus agalactiae: implicationsfor the microbial ‘‘pan-genome.’’ Proc. Natl. Acad. Sci. U. S. A.102:13950–13955.

41. Yang, Y.H., S. Dudoit, P. Luu, D.M. Lin, V. Peng, J. Ngai,

and T.P. Speed. 2002. Normalization for cDNA microarraydata: a robust composite method addressing single andmultiple slide systematic variation. Nucleic Acids Res. 30:e15.

Address correspondence to:Shaynoor Dramsi, Ph.D.

Department of MicrobiologyInstitut Pasteur

Unite de Biologie des Bacteries Pathogenes a Gram-positifParis, F-75015

France

E-mail: shaynoor.dramsi@pasteur.fr

RGA Controls SRR1 AND PILI SYNTHESIS IN GBS 297