Systematic Targeted Mutagenesis of Brucella melitensis 16M Reveals a Major Role for GntR Regulators in the Control of Virulence

Abstract

In order to identify transcriptional regulators involved in virulence gene control in Brucella melitensis, we generated a collection of 88 mutants in the AraC, ArsR, Crp, DeoR, GntR, IclR, LysR, MerR, RpiR, and TetR families of regulators. This collection was named LiMuR (library of mutants for regulators). We developed a method to test several mutants simultaneously in one animal in order to identify those unable to survive. This method, called the plasmid-tagged mutagenesis method, was used to test the residual virulence of mutants after 1 week in a mouse model of infection. Ten attenuated mutants, of which six and three belong to the GntR and LysR families, respectively, were identified and individually confirmed to replicate at lower rates in mice. Among these 10 mutants, only gntR10 and arsR6 are attenuated in cellular models. The LiMuR also allows simple screenings to identify regulators of a particular gene or operon. As a first example, we analyzed the expression of the virB operon in the LiMuR mutants. We carried out Western blottings of whole-cell extracts to analyze the production of VirB proteins using polyclonal antisera against VirB proteins. Four mutants produced small amounts of VirB proteins, and one mutant overexpressed VirB proteins compared to the wild-type strain. In these five mutants, reporter analysis using the virB promoter fused to lacZ showed that three mutants control virB at the transcriptional level. The LiMuR is a resource that will provide straightforward identification of regulators involved in the control of genes of interest.

Bacteria have selected mechanisms to express only the appropriate subset of genes conferring a growth or survival advantage in a given situation (10). The expression of many bacterial genes is regulated at the initiation of transcription by regulators which, in response to specific environmental and/or cellular signals, bind at the promoter of target genes to activate or repress them (23).

The availability of complete bacterial genome sequences has opened up doors to the development of new strategies to study the biology of microorganisms (59). For instance, the complete Brucella melitensis sequence (15) allowed us to conduct a systematic study of several families of regulators in this organism. Bacteria of the genus Brucella are facultative intracellular gram-negative coccobacilli that are pathogenic for domestic animals and occasionally for humans (57). Some transcriptional regulators of Brucella that have a role in the control of virulence have already been identified. Indeed, the two-component system BvrR/BvrS controls cell invasion, intracellular survival, and expression of two outer membrane proteins (25, 58). Some transcriptional regulators required for Brucella's virulence were also identified in several screens in cellular models and in BALB/c mice (19, 31, 35, 36). Random screening for attenuated mutants may not be viewed as a comprehensive analysis that would allow the identification of all regulators involved in the virulence of the model tested. We therefore conducted a systematic disruption strategy for genes putatively coding for transcriptional regulators in B. melitensis 16M.

To identify regulators of virulence genes in Brucella, we selected transcriptional regulator families whose members are known to be involved in host-bacteria interactions (AraC, ArsR, Crp/Fnr, DeoR, GntR, IclR, LysR, MerR, RpiR, and TetR families). Sigma factors and the two-component systems are under investigation elsewhere.

Three families, AraC, ArsR, and IclR families, were chosen for their involvement in virulence (9, 44, 64). Since previous data (37, 61) suggested that Brucella and Rhizobiaceae, which are capable of establishing symbiosis with plants, could share common molecular mechanisms for host-bacterium interactions, we also selected members of the GntR and MerR families. Indeed, Sinorhizobium and its relatives are phylogenetically close to Brucella (42). Rhizobia have the ability to infect the roots of leguminous plants, leading to a symbiotic interaction and eliciting the formation of new organs, the nodules. In these organs, bacteria reduce atmospheric nitrogen into ammonia, which is used by the plants as a nitrogen source (62). The RpiR family was selected because a homolog is found in a cluster of genes putatively involved in the rhizopine metabolism. Rhizopine is thought to be a signal molecule produced by Rhizobium when it is residing in the nodules induced in the roots of leguminous plants (62). Two mutants of Brucella with a disrupted gene involved in rhizopine metabolism are attenuated (18, 36), supporting the hypothesis that Brucella and Rhizobium could share common strategies to interact with their host (61). Members of the Crp, LysR, and TetR families of transcriptional regulators are involved in virulence and in symbiosis (26, 33, 51, 53, 65). Some regulators of the DeoR family are involved in sugar metabolism, but many proteins of that family have no experimentally defined function (43, 66). Nevertheless, the regulator AccR represses opine catabolism and conjugal transfer of the Ti (pTIC58) plasmid in Agrobacterium tumefaciens (4).

In this report, we selected 94 transcriptional regulators grouped in 10 families (Table 1). We successfully mutated 88 genes, and we evaluated the virulence of the mutants in mice by a new method based on signature-tagged mutagenesis (STM) screening. We identified 10 regulators required for Brucella during the first week of infection in mice. The analysis of these regulators allowed us to better understand the environment encountered by the bacteria in a mouse model of infection. We also studied the expression of the virB operon in the library of the 88 mutants, leading to the identification of five new regulators of the virB operon.

TABLE 1.

Putative transcriptional regulators studied in this work and estimation of the residual virulence of the corresponding mutants in BALB/c mice

Family and name	ORFa	Accession no.b	Testc	Result		Family and name	ORPa	Accession no.b	Testc	Result
AraC family
AraC1	I1615	AAL52796	PTM	Not att.
AraC2	II1098	AAL54340	PTM	Not att.
AraC3	II0104	AAL53345	PTM	Not att.
AraC4	II0143	AAL53384	PTM	Not att.
AraC5	I1291	AAL52472	2 mice	Not att.
AraC6	II0648	AAL53890	PTM	Not att.
AraC7	I0685	AAL51866	PTM	Not att.
AraC8	II0721	AAL53963	PTM	Not att.
AraC10	II0814	AAL54056	PTM	Not att.
AraC11	I1384	AAL52565	PTM	Not att.
AraC12	II0405	AAL53647	PTM	Not att.
AraC13	I1558	AAL52739	PTM	Not att.
AraC14	II0641	AAL53883	PTM	Not att.
ArsR family
ArsR1	II0810	AAL54052	NM
ArsR2	II0393	AAL53635	NM
ArsR3	II0763	AAL54005	PTM	Not att.
ArsR4	I0150	AAL51332	PTM	Not att.
ArsR5	I0558	AAL51739	PTM	Not att.
ArsR6	I0430	AAL51611	2 mice	Att.
Crp family
Fnr	I1294	AAL52475	2 mice	Not att.
Crp2	I1752	AAL52933	2 mice	Not att.
Crp3	II0854	AAL54096	PTM	Not att.
NnrB	II0966	AAL54208	PTM	Not att.
NnrA	II0986	AAL54228	PTM	Not att.
NarR	II0947	AAL54189	2 mice	Not att.
DeoR family
DeoR1	II1093	AAL54335	PTM	Not att.
DeoR2	II0426	AAL53668	PTM	Not att.
DeoR3	II0436	AAL53678	PTM	Not att.
DeoR4	I1750	AAL52931	PTM	Not att.
DeoR5	I0305	AAL51486	PTM	Not att.
GntR family
GntR1	II0475	AAL53717	PTM	Att.
GntR2	I0305	AAL53999	2 mice	Att.
GntR3	I1373	AAL52554	PTM	Not att.
GntR4	I0169	AAL51351	PTM	Att.
GntR5	I0881	AAL52062	PTM	Att.
GntR6	II1007	AAL54249	2 mice	Not att.
GntR7	II0071	AAL53312	NM
GntR8	I0254	AAL51436	PTM	Not att.
GntR9	II0226	AAL53467	PTM	Not att.
GntR10	II0116	AAL53357	PTM	Att.
GntR11	II0858	AAL54100	PTM	Not att.
GntR12	II0807	AAL54049	PTM	Not att.
GntR13	I0106	AAL51288	PTM	Not att.
GntR14	II0878	AAL54120	PTM	Not att.
GntR15	II0352	AAL53594	PTM	Not att.
GntR16	II0281	AAL53523	PTM	Not att.
GntR17	I0320	AAL51501	2 mice	Att.
GntR18	II1066	AAL54308	NM
GntR19	II0370	AAL53612	PTM	Not att.
GntR21	II0383	AAL53625	2 mice	Not att.
IclR family
IclR1	II0127	AAL53368	PTM	Not att.
IclR2	II1022	AAL54264	2 mice	Not att.
IclR3	I0387	AAL51568	PTM	Not att.
IclR4	II0299	AAL53541	2 mice	Not att.
IclR5	I1717	AAL52898	PTM	Not att.
IclR6	II0219	AAL53460	2 mice	Not att.
IclR7	II0642	AAL53884	PTM	Not att.
LysR family
LysR1	II0526	AAL53768	PTM	Not att.
LysR2	I1598	AAL52779	PTM	Not att.
LysR3	II0820	AAL54062	PTM	Not att.
LysR4	II0894	AAL54136	PTM	Not att.
LysR6	I0896	AAL52077	PTM	Not att.
LysR7	II0639	AAL53881	PTM	Not att.
LysR8	II0576	AAL53818	PTM	Not att.
LysR9	I0116	AAL51298	PTM	Not att.
LysR10	I1885	AAL53066	PTM	Not att.
LysR11	II0345	AAL53587	PTM	Not att.
LysR12	II0390	AAL53632	PTM	Att.
LysR13	I1913	AAL53094	PTM	Att.
LysR14	I0218	AAL51400	2 mice	Not att.
LysR15	II1065	AAL54307	PTM	Not att.
LysR16	II1077	AAL54319	2 mice	Not att.
LysR17	II1135	AAL54377	PTM	Not att.
LysR18	I1573	AAL52754	PTM	Att.
LysR19	II0493	AAL53735	PTM	Not att.
LysR21	I0513	AAL51694	PTM	Not att.
LysR22	I0686	AAL51867	PTM	Not att.
MerR family
MerR1	I0808	AAL51989	PTM	Not att.
MerR2	II0467	AAL53709	PTM	Not att.
MerR4	I1178	AAL52359	NM
MerR5	I0054	AAL51236	PTM	Not att.
MerR6	II1017	AAL54259	PTM	Not att.
MerR7	I1729	AAL52910	NM
RpiR family
RpiR1	II0545	AAL53787	PTM	Not att.
RpiR2	II0573	AAL53815	2 mice	Not att.
RpiR3	II0556	AAL53798	2 mice	Not att.
TetR family
TetR1	I0604	AAL51785	2 mice	Not att.
TetR2	II1117	AAL54359	PTM	Not att.
TetR3	II0804	AAL54046	2 mice	Not att.
TetR4	I0891	AAL52072	PTM	Not att.
TetR5	I0623	AAL51804	2 mice	Not att.
TetR6	I1641	AAL52822	PTM	Not att.
TetR8	I1631	AAL52812	PTM	Not att.
TetR9	I1379	AAL52560	PTM	Not att.

^aORF number in the B. melitensis genome in GenBank (accession no. AE008917 and AE008918); BMEI1615 is indicated as I1615 in this table.

^bORF accession number in GenBank.

^cThe PTM method was tried for all the mutants (except for the lysR14, rpiR3, and arsR6 mutants). Mutants for which the PTM method does not work were used to infect two mice. Abbreviations are as follows: NM, not mutated; Att., attenuated; Not att., not attenuated.

MATERIALS AND METHODS

Search of transcription factors in the Brucella genome.

The search for regulators was done using the B. melitensis database. The helix-turn-helix (HTH) consensus sequences available from the Pfam database (3) were used to scan the whole Brucella genome deduced proteins with an E value lower than 10⁻³ (3).

The remaining predictions are based on homology to transcriptional regulators from other bacteria described in the literature. In this way, we identified 174 potential transcriptional regulators.

The library of mutants for regulators (LiMuR) and B. melitensis database website addresses are as follows: http://urbm59.urbm.fundp.ac.be/∼dharbi/aPAGe/PHP/basic/pCDS/limur.php and http://serine.urbm.fundp.ac.be/∼seqbruce/GENOMES/Brucella_melitensis/, respectively.

Bacterial strains and plasmids.

B. melitensis 16M was obtained from A. Macmillan, Central Veterinary Laboratory, Weybridge, United Kingdom. We selected a spontaneous nalidixic acid (Nal)-resistant (Nal^r) mutant of this strain that was used as the parental strain for all experiments. B. melitensis strains were grown on solid or liquid 2YT medium with appropriate antibiotics. The Escherichia coli strains used in this study were S17-1 (56) and DH10B (Gibco BRL). E. coli strains were grown on Luria-Bertani (LB) medium with appropriate antibiotics. Antibiotics were used at the following concentrations for E. coli and B. melitensis: kanamycin (Kan), 50 μg/ml; chloramphenicol (Cm), 20 μg/ml; Nal, 25 μg/ml. The plasmids used in this study were pSKOriTKan (12), pBBRMCS1lacZ (S. Léonard, unpublished data), and pMR10Cat Gateway (R. Hallez, unpublished data).

Matings.

Matings were performed by mixing equal volumes (100 μl) of liquid cultures of E. coli S17-1 donor cells (optical density at 600 nm of 0.6) and the B. melitensis 16M Nal^r recipient strain (overnight culture) on a 0.22-μm-pore-size filter. The filter was left for 1 h on a 2YT (log/liter yeast extract, 16 g/liter tryptone, 5 g/liter NaCl) plate without antibiotics and then transferred onto a 2YT plate containing Kan and Nal (or Cm and Nal for replicative plasmids). After 1 week of incubation at 37°C, the exconjugates were replicated on a 2YT plate containing Nal and Kan (or a 2YT plate containing Nal and Cm).

Molecular techniques.

DNA manipulations were performed according to standard techniques (2). Restriction enzymes were purchased from Roche, and primers were purchased from Sigma-Aldrich. Internal fragments were initially amplified by PCR from B. melitensis 16M genomic DNA with primers different for each gene, with common 5′-end sequences, that are as follows: Fint (5′-ATCTCTAGA-3′) and Rint (5′-ATCGTCGACCTA-3′) for forward and reverse primers, respectively. These sequences contain an in-frame stop codon, a restriction site (XbaI or SalI), and an EcoRV half-site. High-fidelity PCR was performed using Pwo polymerase (Boehringer Mannheim) or Pfx polymerase (Gibco BRL). The cycling conditions were 2 min at 94°C followed by 30 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 30 s, and 72°C for 6 min. The PCR products were purified using a Concert Rapid PCR purification system (Gibco BRL). The amplified product was then inserted in the EcoRV-digested pSKOriTKan vector in the opposite orientation compared to the Plac promoter to avoid expression of a 3′ fragment of the disrupted coding sequences (CDS) in Brucella. The inserts were sequenced using the dye terminator method (BigDye Terminator kit; Perkin-Elmer) with an ABI 377 sequencer. The construction was introduced into B. melitensis 16M (Nal^r) from E. coli S17-1 by mating. A single crossover then leads to the disruption of the wild-type locus on the chromosome (Fig. 1). Integrative mutants were selected on a medium with kanamycin and nalidixic acid. One mutant was constructed by allelic replacement, with the arsR6 CDS being replaced by a kanamycin resistance cassette.

FIG. 1.

PTM method. (A) The integrative disruption strategy. The black box represents the coding sequence of the target gene. The Kan^r marker is indicated by a hatched box on the plasmid. A homologous recombination (discontinuous line) occurs between the internal fragment of the gene located on the plasmid and the chromosomal copy of the gene. In the screening strategy, the internal fragment of each target gene is amplified by PCR and spotted on a membrane. (B) The PTM method. Mutants are pooled (input pool) and an aliquot is removed for genomic DNA extraction. The pool is then injected into mice, and after 1 week, the spleens are removed and the mutants are recovered (output pool). The tags, which are the junction between the integrated plasmid and the remaining part of the targeted gene, within the input and output pools are amplified, radiolabeled in a PCR, and hybridized on the internal fragments of the membranes. The comparison of the two membranes allows the identification of mutants that are unable to survive in mice.

Complementation vectors were constructed by using the Gateway technique (Invitrogen) (63). The entry vectors were generated previously (17). Destination vector (pMR10GW) bears kanamycin and chloramphenicol resistance markers, and the toxic cassette is flanked by attR1 and attR2 recombinational sites. The LR recombinational cloning procedure was performed as recommended by the manufacturer (Invitrogen).

Southern blot analysis.

The gene disruption was confirmed by Southern blot analysis using a Kan probe. After chromosomal DNA was extracted, it was digested overnight with HindIII and SspI (or HindIII and EcoRI) restriction enzymes, separated by electrophoresis in 1% agarose, and transferred to positively charged nylon membranes. The Kan cassette was excised from pUC4K (Pharmacia) using EcoRI digestion and was labeled by using a RadPrime DNA labeling system (Invitrogen), and it was used as the DNA probe for Southern hybridization under stringent conditions.

The PTM method. (i) Screening of the mutants.

Mutants were grown at 37°C in 200 μl of 2YT medium in 96-well microtiter plates for 24 h. The bacteria were then pooled (9 to 12 mutants/pool), centrifuged at 4,000 × g for 10 min, and resuspended in 2 ml of phosphate-buffered saline (PBS). The bacterial suspension was then diluted to a final concentration of 5 × 10⁵ CFU in 100 μl of PBS. The number of bacteria was counted by plating dilutions on 2YT plates. The bacterial suspension (100 μl) was injected intraperitoneally into 7-week-old male BALB/c mice (n = 2 mice). The remaining part of the suspension was used for genomic DNA preparation. One week after the infection, animals were sacrificed, and the spleens were removed aseptically. For recovery of bacteria, the spleens were homogenized in PBS-0.1% Triton X-100 (Roche), and dilutions were plated on 2YT medium. Plates containing approximately 10⁵ clones were used for DNA extraction. The plasmid-tagged mutagenesis (PTM) method was tested on 85 mutants. The three remaining mutants were used to infect two mice to complete the testing of virulence for all LiMuR mutants.

(ii) Amplification and labeling of DNA tags.

For both the input and the output pools, bacteria from plates containing approximately 10⁵ clones were resuspended in 3 ml PBS and then centrifuged, and genomic DNA was extracted from the pelleted bacteria. Tags were initially amplified by PCR from genomic DNA with the primers Fint (specific to each gene coding for a regulator) and RpSKKan(5′-CCGCTCTAGAACTAGTGGATC-3′). The amplicons were purified with a Concert Rapid PCR purification system (Gibco BRL), and a fraction (approximately 150 ng) was used as a template for a second PCR including [³²P]dCTP to radiolabel the tags. The cycling conditions for both PCRs were as follows: 2 min at 94°C followed by 30 cycles of 94°C for 30 s, 52°C for 30 s, 72°C for 30 s, and 72°C for 6 min. The reactions were performed with Taq polymerase from Biotools.

For dot blots, 100 μl of a 1:10 dilution of the PCR products (approximately 4 ng) was transferred to a positively charged nylon membrane (Hybond N⁺; Amersham) with a dot blot apparatus (Bio-Rad). After a short prehybridization reaction (in 0.25 M Na₂HPO₄, 1 mM EDTA, sodium dodecyl sulfate [10%], and formamide [40%]) of 5 min, the hybridization reaction was performed overnight.

No labelings were detected in the input pool for 17 mutants (of the 85 mutants tested), which may be due to the impossibility to amplify the signature tags in the multiple PCR. We therefore infected two mice with each of these 17 mutants, and 1 week postinfection (p.i.), we evaluated the number of CFU in spleen and compared it to that of a wild-type control, as described below.

Infection in BALB/c mice.

Seven-week-old male BALB/c mice (n = 4 mice) were inoculated intraperitoneally with 0.1 ml of a suspension containing 5 × 10⁵ CFU of each bacterial strain harvested with PBS from a 24-h 2YT liquid culture. One week after infection, mice were sacrificed for spleen collection. The spleens were homogenized in 2 ml of PBS-0.1% Triton X-100 (Roche), and serial dilutions of the homogenates were plated on 2YT agar to determine bacterial counts. For the wild-type train, 5 to 6 log CFU were recovered using this procedure. The arbitrary attenuation cutoff was at 0.8 log.

Infection of macrophages and HeLa cells.

Subconfluent monolayers (10⁵) of murine J774 macrophages or human HeLa cells were inoculated with bacteria diluted to 6 × 10⁶ CFU ml⁻¹ in cell culture medium. Plates were centrifuged for 10 min at 1,000 × g at room temperature and placed in a 5% CO₂ atmosphere at 37°C. After 1 h, wells were washed three times with PBS and incubated for 48 h with cell culture medium supplemented with 50 μg/ml of gentamicin. At the end of infection time, the monolayers were washed three times with PBS and treated for 10 min with 200 μl of 0.1% Triton X-100 (Roche) in PBS. Serial dilutions of the lysates were plated onto 2YT plates for determination of CFU. Each infection was performed in triplicate. We used the wild-type strain as a positive control (we obtained about 5 log CFU at 48 h p.i. in HeLa cells and macrophages) and the vjbR mutant as a control of attenuation in cellular models of infection (14a).

Western blot analysis.

Bacteria were pregrown in rich medium (2YT) with chloramphenicol, diluted to an optical density at 600 nm of 0.05 in 10 ml of 2YT medium, and grown during 24 h (the final optical density at 600 nm was about 1). After inactivation at 80°C for 1 h, bacteria were pelleted and resuspended in PBS at a concentration of 10¹⁰ CFU/ml. Forty microliters of each total cell lysate was subjected to electrophoresis on 12% sodium dodecyl sulfate-polyacrylamide gels and transferred to Hybond-C nitrocellulose membranes (Amersham Pharmacia Biotech) by the semidry transfer technique (2). Immunodetection of proteins in total cell lysates was first performed with polyclonal VirB9 antiserum (diluted 1/2,500), and the mutants showing a difference in VirB9 abundance compared to that of the wild type were retested using a polyclonal VirB8 antiserum (diluted 1/2,500) (49). The monoclonal anti-Omp89 antibody A5310B2 (8) (diluted 1/1,000) was used as a loading control. Bound antibodies were detected by chemiluminescence with peroxidase-conjugated secondary antibodies and the ECL Western blotting reagent RPN2209 as recommended by the manufacturer (Amersham Pharmacia Biotech).

β-Galactosidase assay.

The sequence located upstream of the virB1 coding sequence (including the ATG itself) was amplified by PCR from genomic DNA of B. melitensis 16M (Nal^r) with the primers FpvirB (5′-TCTAGAGCTAGCTGAAATCCAGGCGTT-3′) and RpvirB (5′-GGATCCACCATAGGATCGTCTCCTTCTCA-3′) containing the XbaI and BamHI restriction sites, respectively. The PCR product was cloned into the pGEM-T easy vector for sequencing. This construct was digested by XbaI and BamHI, and the insert was cloned into the corresponding sites of the broad-host-range plasmid pBBRMCS1lacZ.

Bacteria were pregrown in 2YT medium supplemented with chloramphenicol, diluted to an optical density at 600 nm of 0.05 in 5 ml of 2YT medium (containing chloramphenicol), and grown during 24 h. β-Galactosidase activity (Miller units) was determined on permeabilized bacteria (41).

RESULTS

Identification of putative transcriptional regulator genes.

Prokaryotic transcription factors usually recognize DNA operator sequences using an HTH motif (40). There are several distinct HTH motifs that allow the classification of DNA-binding transcription factors into families (5, 6, 21, 27, 52). First, we searched homologs of HTH consensus sequences available from the Pfam database (3) in B. melitensis 16M genome deduced proteins (seehttp://www.sciences.fundp.ac.be./urbm/IAI_216304_Supp_Material.html). The Pfam database contains conserved domains, many of them having a demonstrated or putative function. We also identified CDS sharing similarities to transcriptional regulators described in other bacteria but not belonging to described families. In this way, we identified 174 CDS putatively encoding transcriptional regulators, with 158 of them distributed among 17 families. The larger families of regulators in B. melitensis are the LysR and GntR families (20 predicted regulators for both families). Thirteen regulators belong to the AraC family, 12 belong to the MarR family, 8 belong to the TetR family, and the remaining regulators belong to the ArsR, Crp, DeoR, IclR, LacI, LuxR, MerR, RpiR, and XRE families. We also identified six CDS homologous to described sigma factors (one RpoD, one RpoN, two RpoH, and two extracytoplasmic factors) and 16 putative response regulators (with a DNA-binding domain) from two-component systems. We selected 10 families (AraC, ArsR, Crp, DeoR, GntR, IclR, LysR, MerR, RpiR, and TetR families) for further studies (Table 1).

Systematic disruption of transcriptional regulators.

We constructed the mutants by integration of a plasmid in the target CDS (integrative disruption) using an internal fragment of the CDS as the site of homologous recombination (Fig. 1A). Each mutant was validated by Southern blot using a probe hybridizing in the kanamycin resistance gene and two restrictions of the genomic DNA. We have obtained 87 integrative mutants of transcriptional regulators. One additional mutant (arsR6) was obtained by replacement of the target CDS with a Kan^r cassette.

The complete collection of mutants, called LiMuR, is available to the scientific community in a 96-well format. Two versions of LiMuR are available, one with the E. coli clones carrying plasmids used for disruption, allowing the targeted mutations in other Brucella strains, and the other being the B. melitensis mutants. A website linked to the B. melitensis database was established to simplify the search for information.

Identification of regulators involved in Brucella's virulence.

In an effort to identify regulators required for Brucella virulence, the ability of LiMuR mutants to survive and replicate in a mouse model of infection was investigated. For obvious bioethic reasons, it was desirable to minimize the number of mice sacrificed for each mutant to test. Thus, we developed a method to test several mutants simultaneously in one animal and then to identify mutants unable to survive. Our strategy is based on STM screening (28), and we called it the PTM method (Fig. 1). We divided the different integrative mutants into nine pools (9 to 12 mutants per pool), and each pool was injected into two mice. This pool size is a compromise between the advantage of mixing many clones and the limitation of multiplex PCR (V. Haine, unpublished data). Considering that each mutant has a plasmid integrated at a specific locus, we used the junction between the disrupted CDS and the integrated plasmid as a tag to differentiate the mutants. In order to check if we are able to identify underrepresented clones, we made preliminary tests by mixing a known amount of mutants in pools, and we determined that the PTM method was able to identify mutants 10- to 100-fold diluted in a pool (data not shown).

First, 9 to 12 mutants were grouped in an input pool (Fig. 1B). Genomic DNA of this pool was extracted, and the tags were amplified and radiolabeled in a PCR. These labeled tags were hybridized on membranes where internal fragments of the target genes without plasmidic sequence are spotted (Fig. 1A). The pools were then injected into mice by the intraperitoneal route. In order to identify regulators involved in the acute phase of infection, mice were killed 1 week p.i., the spleens were recovered and homogenized, and the residual brucellae were allowed to grow on a rich medium, forming the output pool. The tags were also amplified on the output pool, radiolabeled in a PCR, and hybridized on the membranes. The comparison of the input and output membranes allowed the identification of mutants that are unable to survive in mice (Fig. 1B). Like for the STM method, those attenuated mutants present a marked decrease of labeling for the two output pools. The LiMuR mutants were tested by the PTM method, which allowed us to select eight mutants displaying a growth defect in the two mice tested. Twenty signature tags were not amplified from the input pool (see Materials and Methods). We therefore infected two mice for each of these 20 mutants, and 1 week p.i., we evaluated the number of CFU in spleen and compared it to that of a wild-type control. Among these 20 newly tested mutants, we identified 4 attenuated mutants. After this first step of screening, a set of 12 attenuated mutants was therefore identified.

The 12 individual mutants were used to infect groups of four mice intraperitoneally, and the CFU were counted in the spleen after 1 week of infection. We confirmed attenuation for 10 of the 12 selected mutants (Table 2). These 10 mutants display between 0.8 and 2 log of attenuation when expressed as average CFU per spleen. Surprisingly, among these 10 attenuated mutants, 6 mutants belong to the GntR family, which is much more than would be expected by chance (P < 0.01 in a χ² test).

TABLE 2.

Virulence of the mutants in BALB/c mice and in cellular models^a

Strain	Log CFU
	Mice	HeLa	J774
arsR6	2.2 ± 0.1	1.9 ± 0.2	2.2 ± 0.2
gntR1	2.0 ± 0.4	−0.6 ± 0.1	−0.7 ± 0.1
gntR2	1.0 ± 0.1	−0.1 ± 0.1	−0.1 ± 0.1
gntR4	0.9 ± 0.1	0.5 ± 0.1	0.5 ± 0.2
gntR5	0.9 ± 0.1	−0.3 ± 0.1	−0.6 ± 0.3
gntR10	1.2 ± 0.1	0.9 ± 0.1	0.1 ± 0.1
gntR17	0.8 ± 0.2	0.3 ± 0.1	−0.1 ± 0.2
lysR12	1.3 ± 0.1	0.4 ± 0.2	−0.1 ± 0.1
lysR13	0.8 ± 0.2	0.5 ± 0.2	0.3 ± 0.2
lysR18	1.2 ± 0.1	0.7 ± 0.1	0.1 ± 0.1
vjbR	ND	2.0 ± 0.1	1.8 ± 0.2

^aVirulence is expressed as the log of attenuation of the mutant compared to the wild-type strain. The log of attenuation is defined as the log CFU obtained for the wild-type strain minus the log CFU obtained for mutant strain in the same experiment. The vjbR mutant was used as a control of attenuation in cellular models of infection (14a), i.e., HeLa cells and J774 macrophages. ND, not determined.

Since the virulence of Brucella is suspected to be mainly due to its ability to infect and multiply in professional and nonprofessional phagocytes (47), we tested the ability of the 10 attenuated mutants to persist within murine macrophages and HeLa cells. A mutant of the gene coding for the transcriptional regulator VjbR (14a) was used as a control of attenuation. Surprisingly, only one mutant (arsR6) was attenuated in both cellular models, and one mutant (gntR10) was slightly attenuated in HeLa cells only (Table 2).

Implication of the mutants in VirB expression.

In principle, the availability of the LiMuR allows simple screenings for the identification of factors involved in the control of a particular gene or operon. For example, an abundance of a given protein or activity of a selected promoter could be individually evaluated for each mutant of the library.

Numerous data have highlighted a crucial role of the virB operon in the intracellular survival and multiplication of several Brucella species (14, 45, 55). Moreover, several studies showed the importance of this operon in B. abortus pathogenicity in mice (29, 55). In an effort to evaluate the use of LiMuR to identify regulators of a gene of interest, we analyzed virB expression in the 88 mutants available in LiMuR.

We first carried out a Western blotting of whole-cell extracts to investigate whether the mutants produced VirB proteins by using VirB8 and VirB9 polyclonal antisera generated in rabbits (49). We performed a two-step screening in which the abundance of VirB9 and VirB8 was tested (see Materials and Methods). Five mutants showed a difference in VirB8 production compared to the parental strain (Fig. 2). Indeed, VirB8 was either absent or weakly produced in four mutants (arsR6, araC8, deoR1, and gntR4) and overexpressed in one mutant (araC2).

FIG. 2.

Comparison of VirB8 production by the wild-type (wt) strain and mutants for regulator coding genes. The vjbR mutant is used as a negative control (14a). There are 4 × 10⁸ bacteria per well. VirB8 production was evaluated in total cell lysates by Western blotting with VirB8-specific antiserum. The anti-Omp89 antibody was used for the loading control (upper panel).

We tested the implication of these five regulators in the control of the virB operon at the transcriptional level. A translational fusion of the promoter of the virB operon to lacZ was constructed and transferred into the five mutants selected by Western blot analysis and in the wild-type strain. In this reporter system, 496 bp upstream, the B. melitensis virB1 ATG (including the ATG itself) was inserted in fusion with lacZ on a pBBR1-MCS1 plasmid. Expression of the lacZ reporter was determined using β-galactosidase activity assays with cells grown during 24 h in rich medium at a culture phase at which VirB proteins are detectable in B. melitensis extracts. The virB::lacZ expression was twofold less in the vjbR mutant (used as a control) than in the wild-type strain (Fig. 3). Levels of β-galactosidase activity in three mutants (araC8, deoR1, and gntR4) were similar to those detected in the vjbR mutant. The promoter activity of the virB operon was less affected in the arsR6 mutant. Moreover, no overexpression of the virB::lacZ fusion was observed in the araC2 mutant in comparison with the wild-type.

FIG. 3.

β-Galactosidase activity of the virB promoter-lacZ fusion in the wild-type strain, in the vjbR mutant (used as a control of reduced virB expression), and in the five mutants showing a difference in VirB8 production compared to the wild type in Western blotting. β-Galactosidase activity (in Miller units) is the average of three independent clones.

We made complementation experiments to determine whether the low production of VirB proteins observed in the arsR6 and gntR4 mutants (which are attenuated in our screen in mice) was caused by the disruption of the genes coding for those regulators. Complementation of these two mutants with an intact plasmidic copy of the CDS showed wild-type levels of VirB9 production (data not shown).

DISCUSSION

The purpose of this study was to identify regulators required for the virulence of Brucella. The complete genomic sequence of B. melitensis 16M allowed us to search and to classify the 174 potential regulators into families. We conducted a systematic disruption of genes coding for regulator families that are suspected to comprise members involved in host-pathogen interactions. We selected 10 families, which represent 94 transcriptional regulators, and we constructed 88 mutants. The residual virulence of these mutants was evaluated in the murine BALB/c model of infection. The number of animals used was greatly decreased by the development of the PTM method, in which 9 to 12 mutants are simultaneously tested in one animal. We identified 10 regulators required for Brucella's virulence during the first week of infection in BALB/c mice. The genes identified in this study have not been described previously as important for the virulence of Brucella. The analysis of the 10 mutated loci indicated that in only two of the loci (in arsR6 and gntR10), disruption of the targeted CDS may have polar effects on downstream genes in the same operon. Indeed, one mutation is disrupting the last CDS of an operon (gntR2), and the seven other disruptions are located in isolated genes, not in operons, according to the definition given previously by Salgado and colleagues (50). The CDS downstream of arsR6 is a conserved putative RNA methylase homologous to RlmB from E. coli (39), which is a 23S rRNA methyltransferase. Downstream of gntR10, two CDS putatively coding for conserved membrane proteins of unknown function are found. The three CDS located downstream of arsR6 or gntR10 have never been identified as disrupted in previous screens for attenuated mutants (13).

Among the 10 attenuated mutants identified here, six disrupted genes belong to the GntR family and three belong to the LysR family. The GntR family of regulators currently comprises about 300 members that are involved in the regulation of many different biological processes. This family is overrepresented in Brucella compared to Sinorhizobium meliloti, in which only the LysR family predominates (20). A transcriptional regulator of the GntR family is described to be involved in Rhizobium meliloti symbiosis (MocR) (48). A mutant in a gene coding for a GntR-like regulator (GntR18, which was not mutated in this work) was found to be attenuated in an STM screen of B. melitensis in BALB/c mice (36). LysR-type transcriptional regulators (LTTRs) are widely distributed in diverse genera of prokaryotes (52). Most LTTRs are transcriptional activators, but they usually negatively regulate their own expression (52). A member of this family (LysR21) was shown to be required for Brucella pathogenesis in human macrophages (19), but according to our results, a lysR21 mutant is not attenuated after 1 week of infection in mice. Many virulence factors are regulated by LTTRs, and it is not surprising that some regulators required for the virulence of Brucella belong to the LysR family (16, 22, 24, 26).

It was surprising that among the 10 mutants attenuated in BALB/c mice, only two were attenuated in their virulence in cellular models since it was reported that most mutants identified during an STM screen in this mouse model were also attenuated in cellular models (35, 36). We propose that this discrepancy is linked to the fact that only regulator-coding genes are mutated here. Indeed, if the function of a regulator is to transduce a particular signal present in the mouse but absent in cultured cells, e.g., linked to innate immunity, to modify the expression of virulence-associated genes, then a mutant of this regulator will be attenuated in mice but not in cellular models of infection.

Our PTM screen allowed us to identify a mutant which is strongly attenuated in mice as well as in cellular models (HeLa cells and J774 macrophages). This mutant is disrupted in a gene coding for the transcriptional regulator ArsR6. ArsR6 displays 63% amino acid identity with the Sinorhizobium meliloti NolR protein. NolR is known to repress expression of the nodulation genes. Moreover, a proteomic study showed that nolR encodes a global regulatory protein (7, 32). In response to environmental stresses, NolR regulates diverse factors such as heat shock proteins, protein synthesis, and cell growth. The putative global role of ArsR6 in Brucella could explain that an arsR6 mutant is attenuated in all models tested, since other Brucella mutants affected for responses to various stresses are also attenuated in such models (46, 58, 60).

For the other genes identified here as attenuated in mice, we examined their homologs and their genomic context. GntR1 is weakly homologous to UxuR of Geobacillus stearothermophilus (28% of identity over 65 residues), which represses the uxuAB operon involved in the catabolism of glucuronate (54). Furthermore, in the B. melitensis genome, homologs of these uxuAB genes are close to gntR1 as well as the uxaC gene. The UxuAB and UxaC proteins are involved in the Ashwell pathway, which converts glucuronate into 2-keto-3-deoxygluconate (KDG). KDG is phosphorylated by the KDG kinase to give KDGP (2-keto-3-deoxy-6-phosphogluconate). The pathway then converges with the Entner-Doudoroff pathway, and KDGP is finally cleaved by KDGP aldolase to yield glyceraldehyde-3-phosphate and pyruvate (38). These two latter compounds can enter the pentose phosphate pathway and the citrate cycle, respectively. The requirement of a functional gntR1 gene for survival and/or proliferation in BALB/c mice spleen suggests that the absence of an appropriate control of the Ashwell pathway unsettled the sugar metabolism of the bacteria in mice.

The lysR18 gene is surrounded with genes involved in the catabolism of purines (guanine deaminase and xanthine dehydrogenase) (34). Furthermore, homologs of lysR18 in Mesorhizobium loti and Sinorhizobium meliloti are also next to the orthologous genes required for purines catabolism. It may be suggested that LysR18 is involved in the control of purine catabolism genes. Until now, only the synthesis of purines was shown to be required for the virulence of Brucella in both cellular and murine models (1, 11, 30, 31).

Since the LiMuR mutants constructed are useful to study the regulation of a particular gene, we studied the regulation of the virB operon as an example. The virB operon is known to be crucial for the pathogenicity of Brucella (14, 29, 45, 55).

We first carried out Western blottings to analyze the production of VirB proteins using polyclonal antisera against B. suis VirB9 and VirB8. Expression of the VirB8 and VirB9 proteins is reduced in four mutants (araC8, arsR6, deoR1, and gntR4) and overproduced in one mutant (araC2) in comparison with the parental strain during growth in bacteriological medium. The arsR6 and araC2 mutations have only a slight effect on PvirB activity, and moreover, the araC2 mutation is shown to induce higher amounts of VirB8 and VirB9. Additional data are required to test the role of ArsR6 and AraC2 regulators of the virB promoter.

We propose that GntR4, DeoR1, and AraC8 are direct or indirect activators of virB transcription. However, it may not be excluded that these regulators, as AraC2 and ArsR6, may also act at another, i.e., posttranscriptional, regulation level. The GntR4 transcriptional factor was also selected in our screen as required for the pathogenesis of Brucella in mice after 1 week of infection. A B. suis deoR1 mutant is strongly attenuated in human macrophages (31), which is consistent with a possible role of deoR1 in virB activation in this cellular model. Until now, the AraC8 regulator was never described as important for the virulence of Brucella. Additional data are required to investigate the role of AraC8 in virulence control.

It will be interesting to investigate possible links between AraC8, DeoR1, and GntR4 and other regulators of the virB operon, such as the quorum-sensing regulator VjbR, which is an important activator of VirB (14a), or the integration factor (IHF), which was recently shown to interact directly with the virB promoter in B. abortus (55a). Furthermore, the increasing number of regulators involved in virB control suggests the presence of a network of regulators. The PTM method developed in this work allows study of the pathogenesis of several mutants simultaneously in one animal. This method could be systematically applied to other bacterial pathogens. The PTM method was successful on 80% of the integrative mutants, but this percentage is similar to the one observed by Hensel et al. in their first STM screen with Salmonella enterica serovar Typhimurium (28). The LiMuR is a resource available to the scientific community that will be an interesting tool for the study of a given process or a given promoter. Indeed, here we only tested the virulence at 1 week of infection in a mouse model, but other virulence models may be tested using a similar strategy, i.e., using the PTM method. In addition, other processes may be studied, such as resistance to H₂O₂ and polymyxin B. The regulation of particular genes or operons may also be studied by using promoter-reporter fusions or antibodies directed against proteins of interest. The LiMuR may also be extended to other regulators. First, there are six mutants that were not obtained in the course of this study, one of them being reluctant to disruption. That demonstrated the very low proportion of essential genes in the regulator families tested. Finally, there are 80 remaining genes putatively coding for transcriptional regulators, and for the sake of completeness of LiMuR, it would be interesting to mutate them. The availability of the B. melitensis ORFeome (i.e., a collection of all predicted CDS of a genome, each inserted in a plasmid) could provide new gene inactivation strategies (17).