JOURNAL OF BACTERIOLOGY, Nov. 2009, p. 6833–6842 Vol. 191, No. 220021-9193/09/$12.00 doi:10.1128/JB.00734-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Identification of Direct Transcriptional Target Genes of ExoS/ChvITwo-Component Signaling in Sinorhizobium meliloti�

Esther J. Chen,1* Robert F. Fisher,2 Virginia M. Perovich,1 Erich A. Sabio,1 and Sharon R. Long2

Department of Biological Science, Center for Applied Biotechnology Studies, College of Natural Sciences and Mathematics,California State University Fullerton, Fullerton, California 92834-6850,1 and Department of Biological Sciences,

Stanford University, 371 Serra Mall, Stanford, California 94305-50202

Received 4 June 2009/Accepted 1 September 2009

The Sinorhizobium meliloti ExoS/ChvI two-component signaling pathway is required for the development ofa nitrogen-fixing symbiosis between S. meliloti and its plant hosts. ExoS/ChvI also has important roles inregulating succinoglycan production, biofilm formation, motility, nutrient utilization, and the viability offree-living bacteria. Previous microarray experiments with an exoS96::Tn5 mutant indicated that ExoS/ChvIinfluences the expression of a few hundred genes, complicating the investigation of which downstream genesrespond directly or indirectly to ExoS/ChvI regulation. To focus our study of ExoS/ChvI transcriptional targetgenes, we performed transcriptional profiling with chvI gain-of-function and reduced-function strains. The chvIgain-of-function strain that we used contains a dominant gain-of-function chvI allele in addition to wild-typechvI. We identified genes that, relative to their expression level in the wild type, are both upregulated in the chvIgain-of-function strain and downregulated in the reduced-function strain or vice versa. Guided by this focusedset of genes, we performed gel mobility shift assays and demonstrated that ChvI directly binds the intergenicregions upstream of ropB1, SMb21440, and SMc01580. Furthermore, DNase I footprint analysis of the regionupstream of SMc01580 identified a specific DNA sequence bound by ChvI and allowed the discovery of apossible motif for ChvI binding. Our results provide insight into the mechanism of how ExoS/ChvI regulatesits downstream targets and lay a foundation for studying this conserved pathway with critical roles infree-living and symbiotic bacteria.

During symbiosis, root-associated bacteria (rhizobia) supplylegume plants with the fixed nitrogen necessary for plantgrowth in exchange for carbon from the plant. These nitrogen-fixing symbioses result in the formation of root nodules thatenable plant growth in nitrogen-poor soil (2). The earliest stepsin the development of nitrogen-fixing nodules involve commu-nication via secreted chemical signals. Plant flavonoids inducethe rhizobia to synthesize Nod factor, which induces root haircurling and root cortical cell division. Rhizobia colonize curledroot hairs and invade via infection threads that extend throughroot hairs into cells within the nodule. The rhizobia are thenreleased into the cytoplasm of nodule cells in plant membrane-bounded symbiosomes, where they differentiate into nitrogen-fixing bacteroids (21, 30).

Bacterial polysaccharides play an important role in nodu-lation (18). Three polysaccharides that allow Sinorhizobiummeliloti to invade nodules in alfalfa are succinoglycan, galac-toglucan, and K antigen; of these three polysaccharides, suc-cinoglycan is most efficient at mediating infection thread initi-ation and elongation (41). The sequenced S. meliloti strainRm1021 must produce succinoglycan to invade plant roots, sinceRm1021 does not normally produce galactoglucan or symbiot-ically active K antigen (24, 32, 44). Succinoglycan is a polymerof an octasaccharide containing seven glucoses and one galac-

tose, with acetyl, succinyl, and pyruvyl modifications (43). Thesymbiotically active form of succinoglycan is the trimer form(52). Mutants that fail to synthesize succinoglycan or that failto synthesize succinoglycan with the proper modifications havedefects in infection thread initiation and elongation (10).

The ExoS/ChvI two-component system positively regulatesthe transcription of exo genes that encode enzymes for succi-noglycan biosynthesis (11, 14, 54). ExoS is a periplasmic sens-ing histidine kinase that controls the phosphorylation of ChvI,which regulates the transcription of downstream genes (11,36). ExoS/ChvI is negatively regulated by the periplasmic in-hibitor protein ExoR (9, 53). Symbiotic defects can result bothfrom mutations that increase ExoS/ChvI activity [such asexoR95::Tn5, exoR(G76C), or exoS(G268S)] and from muta-tions that decrease ExoS/ChvI activity [such as chvI(K214T)](9, 14, 40, 53, 55).

In addition to its role in succinoglycan synthesis, ExoS/ChvIsignaling is critical for the viability of free-living S. melilotistrains. Attempts to construct null alleles of exoS or chvI in S.meliloti were unsuccessful (11, 39), suggesting that exoS andchvI are essential genes. Besides its roles in viability and sym-biosis, ExoS/ChvI is important for biofilm formation, motility,and nutrient utilization (20, 53, 55). Furthermore, orthologs ofExoS/ChvI in other alphaproteobacteria, BvrS/BvrR in themammalian pathogen Brucella abortus and ChvG/ChvI in theplant pathogen Agrobacterium tumefaciens, are required forvirulence (8, 50).

Despite the importance of ExoS/ChvI signaling in S. meliloti,many of the downstream genes that mediate its various func-tions have not been identified. Transcriptional profiling ofexoR95::Tn5 and exoS96::Tn5 mutants demonstrated that, in

* Corresponding author. Mailing address: Department of BiologicalScience, Center for Applied Biotechnology Studies, College of NaturalSciences and Mathematics, California State University Fullerton, 800N. State College Blvd., Fullerton, CA 92834-6850. Phone: (657) 278-2543. Fax: (657) 278-3426. E-mail: echen@fullerton.edu.

� Published ahead of print on 11 September 2009.

6833

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

addition to exo genes, the expression of hundreds of genes wasaltered (53, 55). The sheer number of potential transcriptionaltargets from these previous studies made it difficult to begin toinvestigate the mechanism of ExoS/ChvI regulation. Genessubject to ExoS/ChvI transcriptional control could not be dis-tinguished easily from genes whose expression was altered asan indirect consequence of other ExoS/ChvI mutant pheno-types. In an attempt to identify new ExoS/ChvI transcriptionaltarget genes, we also tried a genetic screen for suppressorsof the chvI(K214T) mutation, but this screen yielded only ad-ditional mutations in exoR and exoS (9).

To identify a focused set of genes that likely to representtrue transcriptional targets of ExoS/ChvI, we performed mi-croarrays with gain-of-function and reduced-function chvIstrains. The streamlined set of candidate downstream genesrevealed by these studies allowed us to identify direct ExoS/ChvI transcriptional target genes and a binding site for ChvI.

MATERIALS AND METHODS

Strains, media, growth conditions, and genetic techniques. All strains in thisstudy (Table 1) are derived from Rm1021 (streptomycin [Sm]-resistant derivativeof wild-type strain SU47 used for genome sequencing [23]) and were grown at30°C in LB medium. Calcofluor white M2R (Sigma) was filter sterilized andadded to a final concentration of 0.02% in LB agar medium (32). Antibioticswere used at the following concentrations: Sm, 500 �g/ml; neomycin (Nm), 50�g/ml; hygromycin (Hy), 50 �g/ml; spectinomycin (Sp), 50 �g/ml; ampicillin(Ap), 100 �g/ml; kanamycin (Km), 30 �g/ml; and chloramphenicol (Cm), 50�g/ml. All Escherichia coli plasmids were maintained in DH5� cells. Plasmidswere transferred from E. coli to S. meliloti by triparental conjugation using helperplasmid pRK600 (15). N3 phage transduction was performed as described pre-viously (35).

Construction of strains used for transcriptional profiling. The chvI(K214T)(EC69) and wild-type (EC176) strains used for microarray analysis were previ-ously described (53). Both strains are marked with Hyr by the integration ofpDW181 (which is PhisB in pDW33) at hisB, about 10 kb upstream of the chvIlocus. The chvI(D52E)/chvI� strain (EC220) was constructed as follows. Thecomplete chvI open reading frame (ORF), plus 450 bp upstream, was PCRamplified and TA cloned into pCR2.1-TOPO (Invitrogen), generating pEC78.To generate the D52E mutation, site-directed mutagenesis (QuikChange; Strat-agene) of pEC78 was used to replace the GAC codon with GAG at amino acid52, generating pEC97. Both pEC78 and pEC97 were verified by sequencing. TheSpeI/XhoI fragment with the chvI upstream region and the ORF from pEC97was subcloned into the suicide vector pDW33, generating pEC177. pEC177 wasintroduced into Rm1021 by triparental mating, resulting in a strain with bothchvI(D52E) and wild-type chvI. The chvI(D52E)/chvI� allele from the transcon-jugant was transduced into Rm1021 once more, generating EC220.

Construction of uidA transcriptional fusion strains and GUS assays. �-Gluc-uronidase (GUS) fusion plasmids were constructed by PCR amplifying generegions with flanking SpeI/XhoI sites and ligating them into pVO155. The spe-cific regions cloned into each plasmid are as follows: pEC340 contains the 773-bpintergenic region upstream of the exoY ORF, pEC573 contains the 361-bp in-tergenic region upstream of the ropB1 ORF, pEC571 contains the 353-bp inter-genic region upstream of the SMb21188 ORF, pEC616 contains the intergenicregion from 309 bp upstream through the first 91 bp of the SMb21440 ORF,pEC617 contains bp 110 to 442 of the SMb21491 ORF, pEC618 contains 215 bpof the upstream intergenic region and the entire ORF through the stop codon ofSMc01580, pEC565 contains bp 116 to 425 of the SMc01855 ORF, and pEC572contains 191 bp of the intergenic upstream region through the first 160 bp of theSMc00159 ORF. All plasmids were verified by sequencing. None had any mu-tations except pEC565, which had a silent mutation, Gln(CAA3CAG), at aminoacid 58 of the SMc01855 ORF.

Conjugation into Rm1021 resulted in the integration of each transcriptionalfusion plasmid into the S. meliloti genome via a single crossover event. For exoY,ropB1, and SMb21188, the integration event resulted in the duplication of theupstream intergenic region; for SMb21440, SMc01580, and SMc00159, the inte-gration event resulted in the duplication of the upstream region as well as theduplication of all (for SMc01580) or part (for SMb21440 and SMc00159) of the ORF;and for SMb21491 and SMc01855, the integration event resulted in the inter-

ruption of the ORF. Each transcriptional fusion allele was transduced intochvI(K214T) (EC412), chvI(D52E)/chvI� (EC409) and wild-type (EC407)strains. Note that EC69, EC220, and EC176 already contain pDW33-derivedplasmids (with uidA) and thus could not be used for these GUS assays. Twoindependent transductants were each assayed in duplicate for GUS activity aspreviously described (51).

Affymetrix GeneChip analysis. Three biological replicates of each strain werecultured in separate flasks of LB medium with Sm selection to an optical densityat 600 nm of 0.5 to 0.7. RNA was isolated; cDNA was synthesized, labeled, andhybridized; and data were analyzed using Affymetrix software as previouslydescribed (5).

Gel mobility shift assays. Purified recombinant wild-type and D52E ChvIproteins were prepared as follows. N-terminal S-tagged wild-type and S-taggedD52E ChvI were constructed by PCR amplifying full-length wild-type orchvI(D52E) strains with flanking EcoRI/HindIII restriction sites and cloning intopET29a (Novagen), creating pDW231 and pRF1173, respectively. Plasmids were

TABLE 1. Strains and plasmids

Straina orplasmid Genotype or relevant characteristics Reference

or source

StrainsRm1021 Derivative of RCR2011; Sm 37Rm7095 exoR95::Tn5; Sm Nm 14Rm7096 exoS96::Tn5; Sm Nm 14EC69 chvI(K214T) (integrated pDW181); Sm Hy 53EC176 Wild type (integrated pDW181); Sm Hy 53EC220 chvI(D52E)/chvI� (integrated pEC177);

Sm HyThis study

EC407 Wild type (integrated pEC406); Sm Sp This studyEC409 chvI(D52E)/chvI� (integrated pEC405);

Sm SpThis study

EC412 chvI(K214T) (integrated pEC406); Sm Sp 9

PlasmidspDW33 Identical to pVO155; Ap Hy 13pDW181 PhisB in pDW33, Ap Hy 53pEC78 PchvI and chvI in pCR2.1 TOPO; Ap This studypEC97 PchvI and chvI(D52E) in pCR2.1 TOPO;

ApThis study

pEC177 chvI(D52E) in pDW33; Ap Hy This studypEC340 PexoY in pVO155; Ap Nm/Km This studypEC405 chvI(D52E) in pMB439; Ap Sp This studypEC406 PhisB in pMB439; Ap Sp 9pEC565 SMc01855 internal fragment (bp 116-425)

in pVO155; Ap Nm/KmThis study

pEC571 Psmb21188 in pVO155; Ap Nm/Km This studypEC572 Psmc00159 in pVO155; Ap Nm/Km This studypEC573 PropB1 in pVO155; Ap Nm/Km This studypEC616 Psmb21440 in pVO155; Ap Nm/Km This studypEC617 SMb21491 internal fragment (bp 110-442)

in pVO155; Ap Nm/KmThis study

pEC618 Psmc01580 and SMc01580 in pVO155; ApNm/Km

This study

pMB439 pBluescript SK(�) derivative with Spr

cassette; Ap Sp4

pRF1206 Position �359 through 5 bp of ropB1ORF in pCR2.1 TOPO; Ap Km

This study

pRF1211 Position �143 through 11 bp after stopcodon of SMc01580 in pCR2.1 TOPO;Ap Km

This study

pRF1212 Position �297 through start codon ofSMb21440 in pCR2.1 TOPO; Ap Km

This study

pRF1218 Position �390 through �96 before startcodon of SMc01580 in pCR2.1 TOPO;Ap Km

This study

pRK600 Conjugal transfer helper plasmid; Cm 15pVO155 Terminator and polylinker preceding uidA

in pUC119 derivative; Ap Nm/Km38

a All strains are derived from Rm1021.

6834 CHEN ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

verified by sequencing and transformed into BL21(DE3) cells for proteinexpression. S-tagged ChvI proteins were purified using the S-Tag thrombinpurification kit (Novagen). Briefly, cultures were induced with 1 mM IPTG(isopropyl-�-D-thiogalactopyranoside) for 3 h at 37°C, and cells were lysed byfreeze-thaw and sonication. The S-tagged ChvI protein was bound to S-proteinagarose by incubation at 4°C and then cleaved from the agarose beads byincubation with biotinylated thrombin for 90 min at room temperature. Theeluate was concentrated with a BioMax 10-kDa Ultrafree-15 centrifugal filterdevice (Millipore) and stored as aliquots at �80°C in ChvI storage buffer (20 mMHEPES [pH 7.4], 50 mM KCl, 5 mM dithiothreitol, 50% glycerol).

The regions upstream of ropB1, SMb21440, and SMc01580 were 3� end labeledby digesting pRF1206, pRF1211, pRF1212, and pRF1218 with EcoRI (whichflanks the TA-cloned inserts in all four plasmids) and filling in with [�-32P]dATPand unlabeled dCTP, dGTP, and dTTP with the Klenow fragment of DNApolymerase I. The labeled DNA fragments were purified following electrophore-sis on 5% polyacrylamide gels. Any secondary digests to localize the ChvI bindingsite were performed with these gel-purified fragments. Labeled DNA fragmentswere incubated with purified ChvI(D52E) protein at room temperature in bind-ing buffer (10 mM Tris-HCl [pH 7.5], 10 mM KCl, 0.1 mM dithiothreitol, 2.5%glycerol, and 25 �g/ml bovine serum albumin) for 20 min and then run on 5%Tris-borate-EDTA gels at 4°C as described previously (16). Gels were dried, andDNA bands were visualized by autoradiography.

DNase I footprinting. The EcoRI-NruI fragment from pRF1211 and the SalI-EcoRI fragment from pRF1218 were end labeled at the EcoRI ends as describedabove. The labeled fragments were incubated in a final volume of 7 �l withpurified ChvI or ChvI(D52E) protein as described above. A total of 150 mU RQ1DNase (Promega) in a solution containing 25 mM Tris-HCl (pH 7.5), 30 mMMgCl2, and 30 mM CaCl2 was added, and the reaction mixture was incubated for30 s before stopping by the addition of 12 �l DNase stop solution (90% form-amide, 20 mM EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol FF).Samples were heated for 3 min at 95°C, quick cooled on ice, and loaded ontosequencing gels as described previously (17). Maxam-Gilbert DNA sequencingladders (G and G�A) of the same fragments were run next to the DNase Ifootprint samples.

RESULTS

The D52E mutation results in constitutively active chvI.After performing genetic screens to identify new componentsof the ExoS/ChvI pathway and repeatedly isolating suppressormutations in exoR and exoS (9), we decided to pursue agenomic approach. Previously, transcriptional profiling indi-cated that the exoR95::Tn5 and exoS96::Tn5 mutations eachaffected the expression of a few hundred genes (53, 55). Toincrease the stringency of our approach, we chose to performtranscriptional profiling with a chvI partial-loss-of-functionstrain, a chvI gain-of-function strain, and the wild type. Wereasoned that genes that were transcriptionally regulated byExoS/ChvI should show both decreased expression levels in thepartial-loss-of-function strain and increased expression levelsin the gain-of-function strain or vice versa. Conversely, falsepositives and genes that were affected only after physiologicaladaptation to ExoS/ChvI mutant phenotypes were unlikely toshow opposite changes in gene expression in the chvI mutants.In this experiment, we used the chvI(K214T) partial-loss-of-function mutant (EC69) that was described previously (53).

We constructed a strain with increased chvI function throughsite-directed mutagenesis. A mutation of the conserved aspar-tate to glutamate at the site of phosphorylation was previouslyshown to result in the constitutive activation of the NtrC re-sponse regulator (31). We generated the analogous chvI(D52E) mutation and integrated a plasmid with this allele intothe genome of wild-type S. meliloti by a single crossover event.The chvI(D52E) allele behaved like a dominant gain-of-func-tion allele, since this strain (EC220) with one wild-type chvIand one chvI(D52E) allele strongly overproduced succino-

glycan (Fig. 1A). Succinoglycan overproduction in thischvI(D52E)/chvI� strain was due to the increased transcriptionof succinoglycan biosynthesis genes such as exoY (Fig. 1B). Weattempted to replace wild-type chvI with the chvI(D52E) allelevia double recombination but failed to recover strains thatcontained only the chvI(D52E) allele (not shown). We alsotried constructing a constitutively inactive chvI allele by mak-ing a chvI(D52N) mutation, but we were unable to recover anystrains that had wild-type chvI replaced by chvI(D52N) (notshown). These results suggest that the expression of a consti-tutively active or inactive chvI allele on its own may be lethalfor S. meliloti. Thus, we used strain EC220, containing both the

WT

WT (H

yr )

chvI(

K214T

)

chvI(

D52E)/c

hvI+

exoR

95::T

n5

exoS

96::T

n5

0

5

10

15

WT chvI(K214T)

chvI(D52E)/

chvI+

P exoY::GUS

activ

ity(M

iller

uni

ts)

UV

visible

A

B

FIG. 1. chvI(D52E) is a gain-of-function mutation. (A) ThechvI(D52E) mutation results in the overproduction of succinoglycan.Serial dilutions of S. meliloti strains [wild type (WT) (Rm1021), WTmarked with Hyr (EC176), chvI(K214T) (EC69), chvI(D52E)/chvI�

(EC220), exoR95::Tn5 (Rm7095), and exoS96::Tn5 (Rm7096)] werespotted onto LB medium with 0.02% calcofluor plates, incubated at30°C for 3 days, and photographed with UV or visible light. (B) ThechvI(D52E) mutation results in increased exoY transcription. GUSactivity was measured in strains [wild type (EC407), chvI(K214T)(EC412), and chvI(D52E)/chvI� (EC409)] with an integrated PexoY::uidA fusion (pEC340). Two independent transductants were assayed induplicate, and standard deviations are shown. The experiment wasrepeated twice, with similar results.

VOL. 191, 2009 DIRECT TRANSCRIPTIONAL TARGETS OF ExoS/ChvI 6835

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

wild-type and D52E chvI alleles, as the gain-of-function chvIstrain for transcriptional profiling.

Identification of putative chvI target genes via microarrayanalysis. We performed microarray analysis with three S. me-liloti strains: the partial-loss-of-function chvI(K214T) mutant(EC69), the gain-of-function chvI(D52E)/chvI� mutant (EC220),and a wild-type control (EC176). Three biological replicates ofeach strain were cultured in LB medium and analyzed onAffymetrix GeneChips, allowing for nine pairwise compari-sons of gene expression between each mutant and wild-typedata set.

We found 59 genes that showed decreased levels of expres-sion in the chvI(K214T) strain and increased levels of expres-sion in the chvI(D52E)/chvI� strain, or vice versa, compared totheir levels of expression in the wild-type strain (Table 2).Fifty-five of these genes were downregulated in the chvI(K214T) strain and upregulated in the chvI(D52E)/chvI�

strain, suggesting that chvI functions mainly as an activator oftranscription. As a validation of our approach, 15 of the puta-tive ChvI target genes are exo genes involved in succinoglycanproduction, a process known to be regulated by ExoS/ChvI.Also, the transcription of ropB1 appears to be controlled byExoS/ChvI. RopB1 is an outer membrane protein that influ-ences the phage sensitivity of S. meliloti (7), whose orthologs inB. abortus (Omp3A) and A. tumefaciens (AopB) were previ-ously shown to be regulated by ExoS/ChvI orthologs in thoserespective bacteria (27, 33). Only one putative target gene hashomology to a transcriptional regulator, SMa1705 (3). Manyof the putative chvI target genes encode hypothetical proteinswith predicted signal sequences or transmembrane domains(Table 2), according to annotations in the S. meliloti 1021 ge-nome database (http://iant.toulouse.inra.fr/bacteria/annotation/cgi/rhime.cgi) based on the signalp and tmhmm prediction pro-grams.

Other polysaccharide-related genes that appear to be con-trolled by ChvI include the msbA2 operon, eglC, prsD, andopgC. MsbA2 is required for symbiosis, and the msbA2 operon(SMb21188 to SMb21191) is involved in polysaccharide syn-thesis (6, 25). EglC is an endoglycanase that contributes to thedepolymerization of high-molecular-weight succinoglycan (49).PrsD is a component of an ABC transporter that contributesto the secretion of low-molecular-weight succinoglycan (56).OpgC has homology to a cyclic �-1,2-glucan succinyltrans-ferase from Brucella species (45).

Genes involved in inositol catabolism also appear to bepositively regulated by ChvI. The iolB, iolC, iolD, and iolEgenes are predicted to have roles in inositol catabolism basedon homology to genes in Bacillus subtilis (57). Some S. melilotistrains can catabolize rhizopines using the myo-inositol cata-bolic pathway (22), but most S. meliloti strains, includingRm1021, lack the ability to synthesize and catabolize rhizo-pines (46). For Rm1021, the role of inositol catabolism insymbiosis is unclear, but defects in inositol utilization pathwayswere previously shown to decrease the symbiotic competitive-ness of Rhizobium leguminosarum bv. viciae (19) and Sinorhi-zobium fredii (29) and were suggested to decrease the symbi-otic competitiveness of an engineered S. meliloti strain in afield study (48).

Verification of putative chvI target genes. To confirm resultsfrom transcriptional profiling, we made GUS transcriptional

fusions for seven genes in Table 2. The seven genes that wetested were ropB1 and SMb21188 (the first gene in the operonwith msbA2) as well as some of the hypothetical genes with thehighest average signal log ratios (SLRs) in Table 2 (SMb21440,SMb21491, SMc01580, SMc01855, and SMc00159). Each tran-scriptional fusion was integrated into the S. meliloti genome bya single crossover event and transduced into the wild-type,chvI(K214T), and chvI(D52E)/chvI� strains to measure GUSactivity.

GUS activity assays with these strains confirmed the mi-croarray results for these seven genes (Fig. 2). GUS fusions tothe genes that appeared to be positively regulated by ChvI inthe microarray showed lower levels of expression in thechvI(K214T) background and higher levels of expression in thechvI(D52E)/chvI� background than in the wild type. Con-versely, the GUS fusion to SMc00159, which appeared to benegatively regulated by ChvI in the microarray, showed higherlevels of expression in the chvI(K214T) strain and lower levelsof expression in the chvI(D52E)/chvI� strain than in the wild-type strain. These data validate the results of the microarrayexperiment and indicate that ropB1, SMb21188, SMb21440,SMb21491, SMc01580, SMc01855, and SMc00159 are indeedtranscriptionally regulated by ChvI.

Identification of direct chvI transcriptional target genes. Tounderstand how ExoS/ChvI regulates gene expression, wewanted to identify genes whose transcription is directly con-trolled by ChvI binding to DNA. We used gel mobility shiftassays to test 19 upstream regions that correspond to 31 of the59 genes (Table 2), some of which are in operons. We testedupstream regions for most of the exo genes, the seven genesconfirmed to be ChvI regulated in Fig. 2, and some genes fromTable 2 with the highest average SLRs.

We used purified recombinant ChvI(D52E) for the gel shiftassays, reasoning that this mutant protein might bind DNAmore tightly in vitro. Of the 19 upstream regions tested, thoseupstream of ropB1, SMb21440, and SMc01580 showed a mo-bility shift after incubation with the ChvI(D52E) protein (Fig.3). Since the intergenic region upstream of SMc01580 wasrelatively large, overlapping fragments of that region werecloned into two different plasmids, pRF1218 and pRF1211.Both of these fragments showed a mobility shift (Fig. 3C),suggesting that a sequence bound by ChvI upstream of SMc01580 iswithin the overlapping region. Alternatively, there could betwo independent ChvI binding sites on the inserts in pRF1218and pRF1211.

To refine the localization of the ChvI binding site, we per-formed restriction digestion of the full-length fragments. Gelshift assays with products of the restriction digests allowed usto determine which fragments contained the DNA regionbound by ChvI. From these gel shift assays, we found that ChvIbinding sites are within the 116-bp SspI-Sau3AI fragment up-stream of ropB1, the 97-bp Sau96I-AvaI fragment upstream ofSMb21440, and a 48-bp region in the SalI-NruI fragment up-stream of SMc01580 (Fig. 3). Many of the restriction digestsshown in Fig. 3 are partial digests, so the full-length fragmentis still present. The presence of the full-length fragment servesas an internal control for the gel shift assays; had neither of thesmaller fragments shifted, it would have implied that the ChvIbinding site was cleaved by the restriction enzyme. In sum, we

6836 CHEN ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

TABLE 2. ChvI-regulated genes from microarray analysis with chvI(D52E)/chvI� and chvI(K214T) strains

Avg SLR (SD)a

Geneb Gene product and descriptionchvI(D52E) chvI(K214T)

1.60 (0.16) �0.85 (0.13) SMa1587 EglC; endo-1,3-1,4-�-glycanase0.75 (0.34) �1.09 (0.31) SMa1705 Transcriptional regulator, MucR family1.48 (0.38) �0.96 (0.30) SMb20072 ABC transporter, periplasmic solute binding protein; induced by myo-inositol0.88 (0.32) �0.63 (0.08) SMb20329 ThuA; trehalose catabolism protein0.90 (0.13) �0.89 (0.18) SMb20943 ExoZ; acetyltransferase1.17 (0.23) �1.33 (0.20) SMb20944 ExoQ; Putative polysaccharide polymerase, similar to the Wzy protein1.09 (0.18) �1.41 (0.18) SMb20945 ExoF1; periplasmic protein involved in polysaccharide export1.30 (0.20) �1.36 (0.21) SMb20946 ExoY; exopolysaccharide production protein0.70 (0.20) �0.79 (0.17) SMb20948 ExoU; glucosyltransferase0.80 (0.16) �1.58 (0.35) SMb20949 ExoV; pyruvyltransferase1.49 (0.27) �2.00 (0.29) SMb20954 ExoH; succinyltransferase1.36 (0.26) �1.52 (0.34) SMb20955 ExoK; endo-�-1,3-1,4-glucanase1.19 (0.13) �1.07 (0.16) SMb20956 ExoL; putative glucosyltransferase1.30 (0.14) �1.58 (0.13) SMb20957 ExoA; glucosyltransferase1.40 (0.13) �2.10 (0.41) SMb20958 ExoM; glucosyltransferase1.10 (0.16) �1.77 (0.40) SMb20959 ExoO; glucosyltransferase1.12 (0.22) �2.09 (0.17) SMb20960 ExoN; UDPglucose pyrophosphorylase1.22 (0.20) �1.80 (0.36) SMb20961 ExoP; protein-tyrosine kinase0.93 (0.58) �1.26 (0.27) SMb21026 Hypothetical protein1.39 (0.26) �1.26 (0.16) SMb21188 Putative acyltransferase, possibly surface saccharide-specific acetyltransferase1.41 (0.24) �1.46 (0.22) SMb21189 Putative glycosyltransferase0.79 (0.14) �1.23 (0.19) SMb21190 Putative glycosyltransferase1.32 (0.34) �2.86 (0.99) SMb21191 MsbA2; polysaccharide-exporting ABC transporter, ATP binding and permease2.44 (0.27) �3.22 (0.32) SMb21440 Conserved hypothetical protein, signal peptide0.54 (0.17) �0.70 (0.16) SMb21466 PrsD; ABC transporter, ATP binding component, for low-molecular-wt

succinoglycan biosynthesis0.89 (0.12) �1.28 (0.25) SMb21467 Hypothetical protein1.23 (0.07) �1.27 (0.06) SMb21491 Hypothetical exported protein0.98 (0.18) �1.71 (0.14) SMb21690 ExoW; glucosyltransferase1.11 (0.20) �1.33 (0.15) SMc00062 Hypothetical protein1.11 (0.20) �0.93 (0.06) SMc00070 Conserved hypothetical protein signal peptide0.82 (0.16) �1.34 (0.24) SMc00084 Hypothetical protein0.77 (0.23) �1.80 (0.30) SMc00096 Conserved hypothetical protein, signal peptide0.52 (0.27) �1.02 (0.27) SMc00168 SinI; N-acyl-L-homoserine lactone synthetase0.82 (0.24) �0.88 (0.16) SMc00191 Putative signal transduction histidine kinase, phosphotransfer (Hpt) region0.81 (0.23) �2.18 (0.15) SMc00404 Hypothetical protein1.17 (0.15) �1.01 (0.10) SMc00432 IolB; putative myo-inositol catabolism protein1.10 (0.22) �0.95 (0.23) SMc00433 IolE; putative myo-inositol catabolism protein0.60 (0.10) �0.60 (0.05) SMc00604 RopB1; outer membrane protein1.35 (0.20) �1.45 (0.20) SMc01163 Putative oxidoreductase1.22 (0.18) �0.63 (0.31) SMc01165 IolC; putative sugar kinase1.26 (0.07) �0.72 (0.26) SMc01166 IolD; putative malonic semialdehyde oxidative decarboxylase0.85 (0.19) �0.71 (0.12) SMc01556 Hypothetical protein signal peptide2.08 (0.33) �2.76 (0.35) SMc01580 Conserved hypothetical protein1.89 (0.32) �2.49 (0.73) SMc01581 Hypothetical transmembrane protein0.96 (0.38) �1.15 (0.54) SMc01774 Hypothetical protein signal peptide1.72 (0.21) �2.41 (0.20) SMc01855 Putative membrane-bound lytic transglycosylase, signal peptide0.66 (0.11) �0.52 (0.14) SMc02242 Hypothetical transmembrane protein0.77 (0.09) �0.98 (0.15) SMc02317 Hypothetical protein signal peptide0.60 (0.10) �0.67 (0.09) SMc02552 Hypothetical protein1.26 (0.31) �0.63 (0.48) SMc02854 Hypothetical protein signal peptide0.53 (0.13) �0.61 (0.15) SMc02986 Hypothetical transmembrane protein1.13 (0.22) �1.32 (0.23) SMc03108 Calcium binding protein2.10 (0.50) �2.40 (0.88) SMc04236 Putative glycine-rich cell wall structural transmembrane protein1.55 (0.15) �1.22 (0.40) SMc04276 Hypothetical protein signal peptide0.72 (0.10) �0.69 (0.10) SMc04381 OpgC; putative glucan succinyltransferase

�0.60 (0.24) 0.71 (0.32) SMb20902c ABC transporter periplasmic solute binding protein for mannose, L-xylose, glucose,and sorbose

�0.94 (0.44) 0.53 (0.35) SMb21285c Conserved hypothetical protein, transthyretin-like protein�2.10 (0.28) 1.04 (0.31) SMc00159c Conserved hypothetical protein, signal peptide�0.71 (0.11) 0.69 (0.14) SMc00622c Rnd1; RNase D

a Pairwise comparisons between the mutant and wild type yielded average SLR values for each gene. SLR is the log2 ratio of the change, so an SLR of 1 is equivalentto a twofold change. Genes listed here have an average SLR with an absolute value of 0.5 or greater for both the chvI(D52E)/chvI� and chvI(K214T) strains comparedto the wild type and an increase or decrease in expression levels in at least six of the nine pairwise comparisons with the wild type.

b The intergenic upstream regions for the genes with both the gene and gene product in boldface type were tested for binding to ChvI in gel mobility shift assays;those with only the gene product in boldface type appear in an operon whose upstream region was tested for binding to ChvI in gel mobility shift assays.

c Negatively regulated by ChvI. All other genes are positively regulated by ChvI.

VOL. 191, 2009 DIRECT TRANSCRIPTIONAL TARGETS OF ExoS/ChvI 6837

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

identified three genes, ropB1, SMb21440, and SMc01580,whose upstream regions are directly bound by ChvI.

Identification of a DNA sequence bound by ChvI. To pin-point the nucleotide sequence directly bound by ChvI, weperformed DNase I footprint analysis with a region upstreamof SMc01580, the small NruI fragment from pRF1211 (Fig.3C). A 25-bp region corresponding to positions �118 to �94before the start codon of SMc01580 was protected in sampleswith ChvI compared to samples without ChvI (Fig. 4A). Ad-jacent to this region were two sites that were hypersensitive toDNase I cleavage, likely due to a conformational change in-duced by nearby ChvI binding. Footprinting using wild-typerecombinant ChvI and the small SalI fragment of pRF1218(Fig. 3C) radiolabeled on the opposite strand showed essen-tially the same protection patterns (Fig. 4B). Vector sequencesare visible at the bottom of Fig. 4A (right) and B. Figure 4Csummarizes the DNA sequence protected by the direct bindingof ChvI in vitro.

Using the MEME algorithm (http://meme.sdsc.edu), wesearched for a ChvI binding motif in the following 47 se-quences: the footprinted region upstream of SMc01580, thegel-shifted regions upstream of ropB1 and SMb21440, and the44 upstream intergenic sequences (�50 bp in length) corre-sponding to all of the remaining genes in Table 2. We found an11-bp-long motif that was present at least once in each of the47 sequences analyzed, with an overall E value of 8.0e�006(Fig. 5). Visual inspection of the intergenic region upstream ofSMc01580 revealed the presence of another of these motifs;besides the motif within the footprinted region, there is asecond motif about 170 bp upstream of the second one (notshown). Our finding that ChvI binding was detected at only oneof these motifs in gel shift assays indicates that ChvI binding ishighly sequence specific in vitro. In vivo, it is possible thatadditional factors contribute to the specificity of ChvI bindingto DNA.

DISCUSSION

To identify transcriptional target genes of ExoS/ChvI, weused Affymetrix GeneChip microarrays with strains with in-

creased chvI function and reduced chvI function and identified59 putative ExoS/ChvI target genes (including 15 exo genes).Fifty-five of these genes are both upregulated in the chvI gain-of-function strain and downregulated in the chvI reduced-func-tion strain compared to the wild type; the remaining four geneswere downregulated in the gain-of-function strain and upregu-lated in the reduced-function strain. Many of these genes en-code hypothetical proteins with predicted signal sequences ortransmembrane domains. Using transcriptional fusions, weconfirmed that ropB1, SMb21188 (the first gene in the msbA2operon), and five uncharacterized genes are indeed transcrip-tionally regulated by ExoS/ChvI. Furthermore, we identifiedthree direct transcriptional targets of ChvI: ropB1, SMb21440,and SMc01580. Finally, we identified a nucleotide sequencebound by ChvI with DNase I footprinting and found a con-served motif in the intergenic sequences upstream of the 59putative ExoS/ChvI genes. Since ExoS/ChvI is a key signalingpathway in both free-living and symbiotic S. meliloti strains, ourstudy provides an important foundation for future investiga-tions of ExoS/ChvI target genes and the molecular mechanismof ExoS/ChvI regulation.

Previous microarray experiments indicated that the expres-sion of hundreds of genes was altered in exoR95::Tn5 andexoS96::Tn5 mutants (53, 55), making it difficult to begin todistinguish which genes were truly regulated by ExoS/ChvI andwhich genes were affected only indirectly due to physiologicaladaptations to an ExoS/ChvI mutation. Identifying the set ofgenes that were both upregulated in a strain with increasedchvI function and downregulated in a strain with reduced chvIfunction compared to the wild type, or vice versa, allowed us tofocus on a shorter list of genes that were likely to be trueExoS/ChvI targets and facilitated our search for direct ExoS/ChvI transcriptional target genes (Fig. 3). Furthermore, we cannow begin to characterize these genes with respect to the manyfunctions regulated by ExoS/ChvI.

Due to the stringency of our approach, the 59 genes shownin Table 2 likely represent an underestimate of the true num-ber of genes transcriptionally regulated by ExoS/ChvI. Fromtheir genomic context, five of the genes in Table 2 (prsD, thuA,

WT chvI(K214T)chvI(D52E)/chvI+

0

4

8

12

16

GU

S ac

tivity

(nor

mal

ized

to W

T)

ropB1 SMb21188 SMb21440 SMb21491 SMc01580 SMc01855 SMc00159

FIG. 2. ChvI regulates the transcription of ropB1, SMb21188, SMb21440, SMb21491, SMc01580, SMc01855, and SMc00159. GUS activity wasmeasured in strains [wild type (WT) (EC407), chvI(K214T) (EC412), and chvI(D52E)/chvI� (EC409)] with an integrated uidA fusion to eachindicated gene (ropB1 [pEC573], SMb21188 [pEC571], SMb21440 [pEC616], SMb21491 [pEC617], SMc01580 [pEC618], SMc01855 [pEC565], andSMc00159 [pEC572]). Two independent transductants were assayed in duplicate, and standard deviations are shown. The experiment was repeatedtwice, with similar results. The GUS activity of mutant strains was normalized to the GUS activity of the wild type. The absolute GUS activity inMiller units for each fusion in the wild type was as follows: ropB1, 1.86 � 0.04; SMb21188, 0.118 � 0.003; SMb21440, 1.77 � 0.17; SMb21491,0.110 � 0.002; SMc01580, 0.32 � 0.01; SMc01855, 12.89 � 0.16; and SMc00159, 0.45 � 0.06.

6838 CHEN ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

SMc02986, SMb21285, and SMc00159) clearly appear to be-long to operons. Although the other genes in these operonsappear to be coregulated to some extent, their average SLRs orpercentages of pairwise comparisons with significant changesin expression did not meet the cutoff values that we selected forinclusion in Table 2 (not shown). Similarly, we cannot rule outthe possibility that more than 3 of the 19 intergenic regionstested in the gel shift assays are direct targets of ChvI bindingin vivo. For example, we were surprised that we did not detectdirect ChvI binding to exo gene upstream regions in our gelshift assays. However, previous deletion analyses of the exoYupstream region indicated that its transcriptional regulation iscomplex, involving distal and proximal promoter regions andperhaps additional regulatory sites (12, 42). Moreover, it ispossible that our in vitro conditions for ChvI binding in the gelshift assay are more stringent than conditions in vivo.

One group of genes conspicuously absent from Table 2 isgenes involved in chemotaxis and flagellar motility, which were

among the most strongly downregulated genes in previous mi-croarray experiments with exoR95::Tn5 and exoS96::Tn5 (47,53, 55). Since the level of expression of these genes wasstrongly decreased in the chvI(D52E)/chvI� strain, their ab-sence from Table 2 is due to their failure to show increasedexpression levels in the chvI(K214T) mutant compared to wild-type levels. One possible explanation is that flagellar motilitygenes may be expressed at maximum levels in wild-type S.meliloti, so regulatory mechanisms to increase their expressionin a chvI partial-loss-of-function mutant may simply not exist.An alternative explanation is that the downregulation of mo-tility genes in exoR95::Tn5, exoS96::Tn5, and chvI(D52E)/chvI� may be an indirect physiological consequence of ele-vated ExoS/ChvI activity. A ChvI-dependent regulator ofmotility is not evident in Table 2. Although MucR has beenshown to inhibit the transcription of the rem motility regulatorby binding to sequences upstream of rem (1), the MucR familytranscriptional regulator SMa1705 (Table 2) lacks the C2H2

FIG. 3. ChvI binds to DNA upstream of ropB1, SMb21440, and SMc01580. Gel mobility shift assays were performed with purified recombinantChvI(D52E) protein (40 ng) and the upstream intergenic regions of ropB1 (A), SMb21440 (B), and SMc01580 (C). Gel shift assays were firstperformed with the indicated full-length intergenic regions (for A and B) or full-length plasmid inserts (for C). As described in Materials andMethods, these full-length inserts were TA cloned with flanking EcoRI sites, which were used to end label the fragments. Gel shifts were alsoperformed with fragments digested with the indicated restriction enzymes to narrow down the DNA region bound by ChvI. Below each genediagram, thick black lines indicate a fragment whose mobility shifted upon the addition of the ChvI(D52E) protein, and thick gray lines indicatea fragment that was not shifted. Arrowheads indicate a DNA fragment whose mobility shifted upon the addition of ChvI(D52E). Note that manyof the restriction digests are partial digests so that some full-length fragment is still present, serving as an internal positive control for a mobilityshift. Asterisks indicate fragments used for footprinting in Fig. 4.

VOL. 191, 2009 DIRECT TRANSCRIPTIONAL TARGETS OF ExoS/ChvI 6839

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

FIG. 4. ChvI protects a region upstream of SMc01580 from cleavage during DNase I footprint analysis. (A) The small NruI fragment frompRF1211 that showed a mobility shift in Fig. 3C was radiolabeled on the lower strand at its 3� end and subjected to DNase I footprint analysis withpurified ChvI(D52E) protein (100 ng). The left and right panels have the identical samples loaded, but those in the left panel were run longer thanthe samples in the right panel. Thus, much of the same protected sequence at the bottom of the left panel can also be seen at the top of the rightpanel. The black line to the right of each autoradiograph indicates the region protected by ChvI, and asterisks indicate hypersensitive cleavage sites.The gray line indicates a vector sequence from pRF1211. Lanes 1 and 2, Maxam-Gilbert sequencing reactions for G or G�A; lanes 3 and 6, noChvI added; lanes 4 and 5, ChvI(D52E) added; lane 7, no DNase I added. (B) The small SalI fragment from pRF1218 that showed a mobility shiftin Fig. 3C was radiolabeled on the upper strand at its 3� end and subjected to DNase I footprint analysis with purified ChvI(D52E) or wild-type(WT) ChvI protein. The black line indicates the SMc01580 upstream region protected by ChvI, and the gray line indicates the vector sequence frompRF1218. Lanes 1, 2, and 10, Maxam-Gilbert sequencing reactions for G or G�A; lanes 3, 6, and 9, no ChvI added; lanes 4 and 5, ChvI(D52E)added; lanes 7 and 8, ChvI (wild type) added. (C) Sequence of the upstream region of SMc01580 used in the footprint analysis shown in A. Linesindicate the protected regions from A and B, and asterisks indicate hypersensitive cleavage sites. Numbering is with respect to the start codon ofSMc01580.

6840 CHEN ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

zinc finger DNA binding motif found in MucR (3). Also, sinI(Table 2) has been shown to regulate flagellar motility in anexpR� strain background, but no link has been found betweensinI and motility in an expR mutant strain such as Rm1021(26, 28).

Through in silico analyses, we identified a possible ChvIbinding motif (Fig. 5) that is present at least once in all of thesequences upstream of the ChvI-regulated genes listed in Ta-ble 2. This motif is found sometimes on the plus strand andsometimes on the minus strand, and the distances between themotif and the start codon varied from gene to gene. Whetherthe motifs are found at a uniform distance from the transcrip-tion start sites awaits the determination of the transcriptionalstart for each of these genes. For ropB1, the transcriptionalstart site was previously identified (34). The SspI-Sau3AI re-gion that contains the motif and binds to ChvI in gel shiftassays (Fig. 3A) spans from 16 to 124 bp upstream of thepreviously reported ropB1 transcriptional start site.

ExoS/ChvI is a signaling pathway of central importance in S.meliloti, with roles in symbiosis, free-living viability, succino-glycan production, biofilm formation, motility, and nutrientutilization (53, 55). By identifying new ExoS/ChvI transcrip-tional target genes, including three direct transcriptional targetgenes, and a DNA sequence directly bound by ChvI, our studyprovides an important foundation for elucidating how ExoS/ChvI regulates its many downstream functions.

ACKNOWLEDGMENTS

We are grateful to M. Barnett and C. Toman for their generosity andguidance with Affymetrix GeneChip microarrays and analysis, to N.Nikolaidis and D. Wang for assistance with bioinformatics searches,and to M. Benam for construction of the SMc01855::uidA transcrip-tional fusion plasmid. We thank M. Barnett, J. Griffitts, and C. Haneyfor critical reading of the manuscript.

This work was supported by NSF award IOS-0818981 (to E.J.C.), aCalifornia State University Program in Education and Research inBiotechnology faculty-student seed grant (to E.J.C.), postdoctoral fel-lowship PF0507301MBC from the American Cancer Society (toE.J.C.), and NIH award R01-GM30692 (to S.R.L.).

REFERENCES

1. Bahlawane, C., M. McIntosh, E. Krol, and A. Becker. 2008. Sinorhizobiummeliloti regulator MucR couples exopolysaccharide synthesis and motility.Mol. Plant-Microbe Interact. 21:1498–1509.

2. Barnett, M. J., and R. F. Fisher. 2006. Global gene expression in the rhizo-bial-legume symbiosis. Symbiosis 42:1–24.

3. Barnett, M. J., and S. R. Long. 1997. Identification and characterization ofa gene on Rhizobium meliloti pSyma, syrB, that negatively affects syrM ex-pression. Mol. Plant-Microbe Interact. 10:550–559.

4. Barnett, M. J., V. Oke, and S. R. Long. 2000. New genetic tools for use in theRhizobiaceae and other bacteria. BioTechniques 29:240–242, 244–245.

5. Barnett, M. J., C. J. Toman, R. F. Fisher, and S. R. Long. 2004. A dual-genome symbiosis chip for coordinate study of signal exchange and devel-opment in a prokaryote-host interaction. Proc. Natl. Acad. Sci. USA 101:16636–16641.

6. Beck, S., V. L. Marlow, K. Woodall, W. T. Doerrler, E. K. James, and G. P.Ferguson. 2008. The Sinorhizobium meliloti MsbA2 protein is essential forthe legume symbiosis. Microbiology 154:1258–1270.

7. Campbell, G. R., L. A. Sharypova, H. Scheidle, K. M. Jones, K. Niehaus, A.Becker, and G. C. Walker. 2003. Striking complexity of lipopolysaccharidedefects in a collection of Sinorhizobium meliloti mutants. J. Bacteriol. 185:3853–3862.

8. Charles, T. C., and E. W. Nester. 1993. A chromosomally encoded two-component sensory transduction system is required for virulence of Agrobac-terium tumefaciens. J. Bacteriol. 175:6614–6625.

9. Chen, E. J., E. A. Sabio, and S. R. Long. 2008. The periplasmic regulatorExoR inhibits ExoS/ChvI two-component signalling in Sinorhizobium me-liloti. Mol. Microbiol. 69:1290–1303.

10. Cheng, H. P., and G. C. Walker. 1998. Succinoglycan is required for initiationand elongation of infection threads during nodulation of alfalfa by Rhizo-bium meliloti. J. Bacteriol. 180:5183–5191.

11. Cheng, H. P., and G. C. Walker. 1998. Succinoglycan production by Rhizo-bium meliloti is regulated through the ExoS-ChvI two-component regulatorysystem. J. Bacteriol. 180:20–26.

12. Cheng, H. P., and S. Y. Yao. 2004. The key Sinorhizobium meliloti succino-glycan biosynthesis gene exoY is expressed from two promoters. FEMS Mi-crobiol. Lett. 231:131–136.

13. Cronan, G. E., and D. H. Keating. 2004. Sinorhizobium meliloti sulfotrans-ferase that modifies lipopolysaccharide. J. Bacteriol. 186:4168–4176.

14. Doherty, D., J. A. Leigh, J. Glazebrook, and G. C. Walker. 1988. Rhizobiummeliloti mutants that overproduce the R. meliloti acidic calcofluor-bindingexopolysaccharide. J. Bacteriol. 170:4249–4256.

15. Finan, T. M., B. Kunkel, G. F. De Vos, and E. R. Signer. 1986. Secondsymbiotic megaplasmid in Rhizobium meliloti carrying exopolysaccharide andthiamine synthesis genes. J. Bacteriol. 167:66–72.

16. Fisher, R. F., T. T. Egelhoff, J. T. Mulligan, and S. R. Long. 1988. Specificbinding of proteins from Rhizobium meliloti cell-free extracts containingNodD to DNA sequences upstream of inducible nodulation genes. GenesDev. 2:282–293.

17. Fisher, R. F., and S. R. Long. 1989. DNA footprint analysis of the transcrip-tional activator proteins NodD1 and NodD3 on inducible nod gene promot-ers. J. Bacteriol. 171:5492–5502.

18. Fraysse, N., F. Couderc, and V. Poinsot. 2003. Surface polysaccharide in-volvement in establishing the rhizobium-legume symbiosis. Eur. J. Biochem.270:1365–1380.

19. Fry, J., M. Wood, and P. S. Poole. 2001. Investigation of myo-inositol catab-olism in Rhizobium leguminosarum bv. viciae and its effect on nodulationcompetitiveness. Mol. Plant-Microbe Interact. 14:1016–1025.

20. Fujishige, N. A., N. N. Kapadia, P. L. De Hoff, and A. M. Hirsch. 2006.Investigations of Rhizobium biofilm formation. FEMS Microbiol. Ecol. 56:195–206.

21. Gage, D. J. 2004. Infection and invasion of roots by symbiotic, nitrogen-fixingrhizobia during nodulation of temperate legumes. Microbiol. Mol. Biol. Rev.68:280–300.

22. Galbraith, M. P., S. F. Feng, J. Borneman, E. W. Triplett, F. J. de Bruijn,and S. Rossbach. 1998. A functional myo-inositol catabolism pathway isessential for rhizopine utilization by Sinorhizobium meliloti. Microbiology144:2915–2924.

23. Galibert, F., T. M. Finan, S. R. Long, A. Puhler, P. Abola, F. Ampe, F.Barloy-Hubler, M. J. Barnett, A. Becker, P. Boistard, G. Bothe, M. Boutry,L. Bowser, J. Buhrmester, E. Cadieu, D. Capela, P. Chain, A. Cowie, R. W.Davis, S. Dreano, N. A. Federspiel, R. F. Fisher, S. Gloux, T. Godrie, A.Goffeau, B. Golding, J. Gouzy, M. Gurjal, I. Hernandez-Lucas, A. Hong, L.Huizar, R. W. Hyman, T. Jones, D. Kahn, M. L. Kahn, S. Kalman, D. H.Keating, E. Kiss, C. Komp, V. Lelaure, D. Masuy, C. Palm, M. C. Peck, T. M.Pohl, D. Portetelle, B. Purnelle, U. Ramsperger, R. Surzycki, P. Thebault, M.Vandenbol, F. J. Vorholter, S. Weidner, D. H. Wells, K. Wong, K. C. Yeh, andJ. Batut. 2001. The composite genome of the legume symbiont Sinorhizo-bium meliloti. Science 293:668–672.

24. Gonzalez, J. E., B. L. Reuhs, and G. C. Walker. 1996. Low molecular weight

FIG. 5. Motif found in the upstream region of each ChvI-depen-dent gene or operon. The footprinted region upstream of SMc01580,the mobility-shifted regions upstream of ropB1 and SMb21440, and theremaining 44 intergenic regions (�50 bp long) corresponding to eachgene or operon in Table 2 were analyzed using the MEME algorithm.The sequence logo represents a motif that occurs at least once in eachinput sequence; the log likelihood ratio of this motif is 349, and the Evalue is 8.0e�006. The E value is an estimate of the number of11-bp-long motifs with a log likelihood ratio of 349 or higher thatwould be expected to occur once in each input sequence of a similarlysized set of 47 random sequences. SSC, small-sample correction.

VOL. 191, 2009 DIRECT TRANSCRIPTIONAL TARGETS OF ExoS/ChvI 6841

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.

EPS II of Rhizobium meliloti allows nodule invasion in Medicago sativa. Proc.Natl. Acad. Sci. USA 93:8636–8641.

25. Griffitts, J. S., and S. R. Long. 2008. A symbiotic mutant of Sinorhizobiummeliloti reveals a novel genetic pathway involving succinoglycan biosyntheticfunctions. Mol. Microbiol. 67:1292–1306.

26. Gurich, N., and J. E. Gonzalez. 2009. Role of quorum sensing in Sinorhizo-bium meliloti-alfalfa symbiosis. J. Bacteriol. 191:4372–4382.

27. Guzman-Verri, C., L. Manterola, A. Sola-Landa, A. Parra, A. Cloeckaert, J.Garin, J. P. Gorvel, I. Moriyon, E. Moreno, and I. Lopez-Goni. 2002. Thetwo-component system BvrR/BvrS essential for Brucella abortus virulenceregulates the expression of outer membrane proteins with counterparts inmembers of the Rhizobiaceae. Proc. Natl. Acad. Sci. USA 99:12375–12380.

28. Hoang, H. H., A. Becker, and J. E. Gonzalez. 2004. The LuxR homologExpR, in combination with the Sin quorum-sensing system, plays a centralrole in Sinorhizobium meliloti gene expression. J. Bacteriol. 186:5460–5472.

29. Jiang, G., A. H. Krishnan, Y. W. Kim, T. J. Wacek, and H. B. Krishnan. 2001.A functional myo-inositol dehydrogenase gene is required for efficient nitro-gen fixation and competitiveness of Sinorhizobium fredii USDA191 to nod-ulate soybean (Glycine max [L.] Merr.). J. Bacteriol. 183:2595–2604.

30. Jones, K. M., H. Kobayashi, B. W. Davies, M. E. Taga, and G. C. Walker.2007. How rhizobial symbionts invade plants: the Sinorhizobium-Medicagomodel. Nat. Rev. Microbiol. 5:619–633.

31. Klose, K. E., D. S. Weiss, and S. Kustu. 1993. Glutamate at the site ofphosphorylation of nitrogen-regulatory protein NTRC mimics aspartyl-phos-phate and activates the protein. J. Mol. Biol. 232:67–78.

32. Leigh, J. A., E. R. Signer, and G. C. Walker. 1985. Exopolysaccharide-deficient mutants of Rhizobium meliloti that form ineffective nodules. Proc.Natl. Acad. Sci. USA 82:6231–6235.

33. Li, L., Y. Jia, Q. Hou, T. C. Charles, E. W. Nester, and S. Q. Pan. 2002. Aglobal pH sensor: Agrobacterium sensor protein ChvG regulates acid-induc-ible genes on its two chromosomes and Ti plasmid. Proc. Natl. Acad. Sci.USA 99:12369–12374.

34. MacLellan, S. R., A. M. MacLean, and T. M. Finan. 2006. Promoter pre-diction in the rhizobia. Microbiology 152:1751–1763.

35. Martin, M. O., and S. R. Long. 1984. Generalized transduction in Rhizobiummeliloti. J. Bacteriol. 159:125–129.

36. Mascher, T., J. D. Helmann, and G. Unden. 2006. Stimulus perception inbacterial signal-transducing histidine kinases. Microbiol. Mol. Biol. Rev.70:910–938.

37. Meade, H. M., S. R. Long, G. B. Ruvkun, S. E. Brown, and F. M. Ausubel.1982. Physical and genetic characterization of symbiotic and auxotrophicmutants of Rhizobium meliloti induced by transposon Tn5 mutagenesis. J.Bacteriol. 149:114–122.

38. Oke, V., and S. R. Long. 1999. Bacterial genes induced within the noduleduring the Rhizobium-legume symbiosis. Mol. Microbiol. 32:837–849.

39. Østerås, M., J. Stanley, and T. M. Finan. 1995. Identification of Rhizobium-specific intergenic mosaic elements within an essential two-component reg-ulatory system of Rhizobium species. J. Bacteriol. 177:5485–5494.

40. Ozga, D. A., J. C. Lara, and J. A. Leigh. 1994. The regulation of exopolysac-charide production is important at two levels of nodule development inRhizobium meliloti. Mol. Plant-Microbe Interact. 7:758–765.

41. Pellock, B. J., H. P. Cheng, and G. C. Walker. 2000. Alfalfa root noduleinvasion efficiency is dependent on Sinorhizobium meliloti polysaccharides. J.Bacteriol. 182:4310–4318.

42. Quester, I., and A. Becker. 2004. Four promoters subject to regulation by

ExoR and PhoB direct transcription of the Sinorhizobium meliloti exoYFQoperon involved in the biosynthesis of succinoglycan. J. Mol. Microbiol.Biotechnol. 7:115–132.

43. Reinhold, B. B., S. Y. Chan, T. L. Reuber, A. Marra, G. C. Walker, and V. N.Reinhold. 1994. Detailed structural characterization of succinoglycan, themajor exopolysaccharide of Rhizobium meliloti Rm1021. J. Bacteriol. 176:1997–2002.

44. Reuhs, B. L., M. N. Williams, J. S. Kim, R. W. Carlson, and F. Cote. 1995.Suppression of the Fix� phenotype of Rhizobium meliloti exoB mutants bylpsZ is correlated to a modified expression of the K polysaccharide. J.Bacteriol. 177:4289–4296.

45. Roset, M. S., A. E. Ciocchini, R. A. Ugalde, and N. Inon de Iannino. 2006.The Brucella abortus cyclic �-1,2-glucan virulence factor is substituted withO-ester-linked succinyl residues. J. Bacteriol. 188:5003–5013.

46. Rossbach, S., G. Rasul, M. Schneider, B. Eardly, and F. J. de Bruijn. 1995.Structural and functional conservation of the rhizopine catabolism (moc)locus is limited to selected Rhizobium meliloti strains and unrelated to theirgeographical origin. Mol. Plant-Microbe Interact. 8:549–559.

47. Rotter, C., S. Muhlbacher, D. Salamon, R. Schmitt, and B. Scharf. 2006.Rem, a new transcriptional activator of motility and chemotaxis in Sinorhi-zobium meliloti. J. Bacteriol. 188:6932–6942.

48. Scupham, A. J., A. H. Bosworth, W. R. Ellis, T. J. Wacek, K. A. Albrecht, andE. W. Triplett. 1996. Inoculation with Sinorhizobium meliloti RMBPC-2increases alfalfa yield compared with inoculation with a nonengineered wild-type strain. Appl. Environ. Microbiol. 62:4260–4262.

49. Sharypova, L. A., S. N. Yurgel, M. Keller, B. V. Simarov, A. Puhler, and A.Becker. 1999. The eff-482 locus of Sinorhizobium meliloti CXM1-105 thatinfluences symbiotic effectiveness consists of three genes encoding an en-doglycanase, a transcriptional regulator and an adenylate cyclase. Mol. Gen.Genet. 261:1032–1044.

50. Sola-Landa, A., J. Pizarro-Cerda, M. J. Grillo, E. Moreno, I. Moriyon, J. M.Blasco, J. P. Gorvel, and I. Lopez-Goni. 1998. A two-component regulatorysystem playing a critical role in plant pathogens and endosymbionts ispresent in Brucella abortus and controls cell invasion and virulence. Mol.Microbiol. 29:125–138.

51. Swanson, J. A., J. T. Mulligan, and S. R. Long. 1993. Regulation of syrM andnodD3 in Rhizobium meliloti. Genetics 134:435–444.

52. Wang, L. X., Y. Wang, B. Pellock, and G. C. Walker. 1999. Structuralcharacterization of the symbiotically important low-molecular-weight succi-noglycan of Sinorhizobium meliloti. J. Bacteriol. 181:6788–6796.

53. Wells, D. H., E. J. Chen, R. F. Fisher, and S. R. Long. 2007. ExoR isgenetically coupled to the ExoS-ChvI two-component system and located inthe periplasm of Sinorhizobium meliloti. Mol. Microbiol. 64:647–664.

54. Wells, D. H., and S. R. Long. 2002. The Sinorhizobium meliloti stringentresponse affects multiple aspects of symbiosis. Mol. Microbiol. 43:1115–1127.

55. Yao, S. Y., L. Luo, K. J. Har, A. Becker, S. Ruberg, G. Q. Yu, J. B. Zhu, andH. P. Cheng. 2004. Sinorhizobium meliloti ExoR and ExoS proteins regulateboth succinoglycan and flagellum production. J. Bacteriol. 186:6042–6049.

56. York, G. M., and G. C. Walker. 1997. The Rhizobium meliloti exoK gene andprsD/prsE/exsH genes are components of independent degradative pathwayswhich contribute to production of low-molecular-weight succinoglycan. Mol.Microbiol. 25:117–134.

57. Yoshida, K., M. Yamaguchi, T. Morinaga, M. Kinehara, M. Ikeuchi, H.Ashida, and Y. Fujita. 2008. myo-Inositol catabolism in Bacillus subtilis.J. Biol. Chem. 283:10415–10424.

6842 CHEN ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

02

Janu

ary

2022

by

190.

53.8

9.99

.