Genome Expression Analyses Revealing the Modulation of the Salmonella Rcs Regulon by the Attenuator IgaA

Abstract

Intracellular growth attenuator A (IgaA) was identified as a Salmonella enterica regulator limiting bacterial growth inside fibroblasts. Genetic evidence further linked IgaA to repression of the RcsCDB regulatory system, which responds to envelope stress. How IgaA attenuates this system is unknown. Here, we present genome expression profiling data of S. enterica serovar Typhimurium igaA mutants grown at high osmolarity and displaying exacerbated Rcs responses. Transcriptome data revealed that IgaA attenuates gene expression changes requiring phosphorylated RcsB (RcsB∼P) activity. Some RcsB-regulated genes, yciGFE and STM1862 (pagO)-STM1863-STM1864, were equally expressed in wild-type and igaA strains, suggesting a maximal expression at low levels of RcsB∼P. Other genes, such as metB, ypeC, ygaC, glnK, glnP, napA, glpA, and nirB, were shown for the first time and by independent methods to be regulated by the RcsCDB system. Interestingly, IgaA-deficient strains with reduced RcsC or RcsD levels exhibited different Rcs responses and distinct virulence properties. spv virulence genes were differentially expressed in most of the analyzed strains. spvA expression required RcsB and IgaA but, unexpectedly, was also impaired upon stimulation of the RcsC→RcsD→RcsB phosphorelay. Overproduction of either RcsB⁺ or a nonphosphorylatable RcsB(D56Q) variant in strains displaying low spvA expression unveiled that both dephosphorylated RcsB and RcsB∼P are required for optimal spvA expression. Taken together, our data support a model with IgaA attenuating the RcsCDB system by favoring the switch of RcsB∼P to the dephosphorylated state. This role of IgaA in constantly fine-tuning the RcsB∼P/RcsB ratio may ensure the proper expression of important virulence factors, such as the Spv proteins.

Bacteria use two-component signaling systems to sense, respond to, and adapt to changes in the environment (38, 57). These systems participate in processes such as chemotaxis, motility, biofilm formation, adaptation to stress, and virulence (1, 5, 24). A sensor histidine kinase and a cytosolic response regulator are essential elements of these systems (38, 43). In response to the external stimulus, the histidine kinase auto-phosphorylates a conserved histidine residue to further transfer the phosphoryl group to a conserved aspartate residue of the response regulator. Once phosphorylated, the response regulator binds target genes to modulate their expression levels. Transfer of the phosphate group also occurs between modules of the same protein or intermediate proteins that act as phospho-transmitters (43). A widely studied two-component signaling system is the RcsCDB signaling system, first identified in Escherichia coli as a regulator of the production of the colanic acid capsule (53). This system can be activated by diverse envelope-related stresses (13, 14, 19, 20, 35, 36, 47, 49, 52, 61) or the overproduction of proteins such as DjlA, LolA, and OmpG (11, 37).

The RcsCDB system operates with two membrane proteins, the hybrid sensor RcsC and the phospho-transmitter RcsD, together with the cytosolic response regulator RcsB (34, 39). Most genes subjected to regulation by the RcsCDB system encode envelope proteins and factors related to the synthesis of diverse exopolysaccharides (20, 22, 23, 27, 30, 35). It has been proposed that, upon sensing envelope alterations, bacteria activate the RcsCDB system to remodel the envelope composition accordingly. The stimulus sensed by the system is transmitted via an RcsC→RcsD→RcsB phosphorelay (11). Phosphorylated RcsB (RcsB∼P) modulates target gene expression by interacting with DNA through its C-terminal DNA-binding domain. RcsA is another protein playing a coregulatory role in a subset of the RcsB-regulated genes, such as those involved in the production of the colanic acid capsule (39). The outer membrane protein RcsF also modulates the RcsCDB system by activating RcsC via a yet-unknown mechanism (8, 40). RcsF-independent signaling requiring RcsC has also been shown previously (8). Another important regulatory mechanism is the phosphorylation of RcsB mediated by acetyl-phosphate under conditions in which RcsC functions mainly as phosphatase (26).

In our previous studies, we obtained genetic evidence linking an inner membrane protein of Salmonella enterica serovar Typhimurium, named intracellular growth attenuator A (IgaA), to repression of the RcsCDB system (6). Mutations affecting IgaA stability result in strong RcsCDB responses, manifested by the inhibition of flagellum production and the overproduction of capsule material and cell division proteins (6, 42). Activation of the RcsCDB system in E. coli has been correlated in most cases to envelope stress (34, 39). However, mutations in IgaA causing partial loss of function (igaA1 allele) do not alter envelope integrity in serovar Typhimurium (17, 42). Thus, activation of the RcsCDB system does not necessarily impair envelope homeostasis, and some remodeling may be tolerable without compromising basic envelope functions. On the other hand, IgaA is an essential protein in serovar Typhimurium but dispensable in mutants lacking any of the three main components of the RcsCDB system (6, 15, 42). Spontaneous mutations consisting of deletions leading to a diminished production of RcsC or RcsD are also known to suppress IgaA essentiality (42). Important traits of serovar Typhimurium, such as its virulence potential, depend on the fine-tuning of the activity of the RcsCDB system (17, 28, 45). IgaA and the RcsCDB system are present in the same genera of enteric bacteria (17, 20, 34), and this concurrence strongly supports the functional link existing between IgaA and the RcsCDB system.

Microarray analyses performed with E. coli have been instrumental in defining the composition of the Rcs regulon under diverse environmental conditions (23, 30, 46). Although common genes were found in these studies, there were also substantial differences, probably due to the distinct growth conditions and the diverse mutations used. Thus, 46 genes with altered expression were identified when rcsB⁺ and ΔrcsB strains were compared (46); 32 genes were identified when rcsC⁺ and ΔrcsC strains grown at low temperatures, in glucose, and with high zinc concentrations were compared (30); and 149 genes were identified when profiles of rcsC⁺ and ΔrcsC bacteria overexpressing DjlA were compared (23). The Rcs regulon of Yersinia pseudotuberculosis has been estimated to be formed by at least 136 genes, with 60% of the genes encoding functions related to envelope homeostasis (31). In serovar Typhimurium, 26 RcsB-dependent but RcsA-independent genes were reported upon exposure of bacteria to sublethal concentrations of antimicrobial peptides (20). Recent studies performed with serovar Typhimurium expressing IgaA(T191P), a variant previously shown by us to be unable to repress the RcsCDB system (17), revealed that one-fifth of the genes analyzed displayed expression changes higher than twofold (55).

In this work, we analyzed the serovar Typhimurium Rcs regulon in bacteria exposed to high osmolarity, a condition known to stimulate the RcsCDB system (61). We also investigated the Rcs regulon in strains producing an IgaA(R188H) variant (R188H is the first mutation characterized for this attenuator to cause overactivation of the RcsCDB system [6, 7, 17, 27, 54]) and in recently described IgaA-deficient strains having reduced RcsC or RcsD levels (42). The study has provided new insights into the mode of attenuation exerted by IgaA on the RcsCDB system and has led to the first evidence supporting a role for dephosphorylated RcsB as a transcriptional regulator.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The Salmonella enterica serovar Typhimurium strains and plasmids used in this work are listed in Table 1. All strains are isogenic derivates of the mouse-virulent strain SL1344 (33). Bacteria were grown at 37°C in Luria-Bertani (LB) broth, on LB agar plates, or in minimal intracellular-salt medium (ISM) as indicated previously (6, 58). ISM, which is of high osmolarity (403 mosmol), provides an optimal stimulus for induction of the RcsCDB system in strains with defects in IgaA (see below). ISM has also been shown to be suitable to assess expression of the spv virulence genes in serovar Typhimurium (58). Unless otherwise indicated, glycerol (38 mM) was used as the carbon source in the ISM. When appropriate, kanamycin (30 μg ml⁻¹), tetracycline (10 μg ml⁻¹), ampicillin (100 μg ml⁻¹), or chloramphenicol (10 μg ml⁻¹) was used.

TABLE 1.

S. enterica serovar Typhimurium strains and plasmids used in this study

Strain	Relevant genotype	Reference or source
Strains
SL1344	hisG46 rpsL, virulent parental isolate	33
SV4450	igaA1	6
MD0862	igaA1 rcsB70::Tn10d Cm	42
SV4406	rcsB70::Tn10d Cm	6
SV4343	igaA1 zhf-6311::Tn10d Tet rcsA51::MudQ	6
SV4379	rcsA51::MudP	6
SV4380	rcsC52::MudQ	6
SV4404	igaA1 zhf-6311::Tn10d Tet rcsC52::MudQ	6
MD2023	ΔrcsD69	This work
MD2026	igaA1 zhf-6311::Tn10d Tet ΔrcsD69	This work
MD0835-J1	igaA2::KIXX Δ(apbE′-rcsC′)	42
MD0842-J4	igaA2::KIXX rcsC67′	42
MD0855-J10	igaA2::KIXX Δ(ompC′-micF)	42
MD0890-J4	igaA3::CamrrcsC67′	This work
MD1313-J10	igaA3::Camr Δ(ompC′-micF)	This work
SV4407	wcaH21::MudJ	6
MD0099	igaA1 wcaH21::MudJ	6
MD0870	igaA1 rcsB70::Tn10d Cm wcaH21::MudJ	This work
MD0895-J4	igaA3::CamrrcsC67′ wcaH-21::MudJ	42
MD1326-J10	igaA3::Camr Δ(ompC′-micF) wcaH21::MudJ	42
SV4312	spvA103::MudJ	7
MD1368	igaA1 spvA103::MudJ	This work
MD1369	igaA1 rcsB70::Tn10d Cm spvA103::MudJ	This work
MD1370	rcsB70::Tn10d Cm spvA103::MudJ	This work
MD1371-J4	igaA3::CamrrcsC67′ spvA103::MudJ	This work
MD1372-J10	igaA3::Camr Δ(ompC′-micF) spvA103::MudJ	This work
MD1396	igaA1 zhf-6311::Tn10d Tet rcsA51::MudQ spvA103::MudJ	This work
MD1397	rcsA51::MudQ spvA103::MudJ	This work
MD2007	rcsC52::MudQ spvA103::MudJ	This work
MD2008	igaA1 zhf-6311::Tn10d Tet rcsC52::MudQ spvA103::MudJ	This work
MD2027	ΔrcsD69 spvA103::MudJ	This work
MD2028	igaA1 zhf-6311::Tn10d Tet ΔrcsD69 spvA103::MudJ	This work
MD2053	igaA1 rcsB70::Tn10d Cm/pIZ1589	This work
MD2054	igaA1 rcsB70::Tn10d Cm/pJM0029	This work
MD2055	spvA103::MudJ/pIZ1589	This work
MD2056	igaA1 spvA103::MudJ/pIZ1589	This work
MD2057	igaA1 rcsB70::Tn10d Cm spvA103::MudJ/pIZ1589	This work
MD2058	rcsB70::Tn10d Cm spvA103::MudJ/pIZ1589	This work
MD2059	spvA103::MudJ/pJM0029	This work
MD2060	igaA1 spvA103::MudJ/pJM0029	This work
MD2061	igaA1 rcsB70::Tn10d Cm spvA103::MudJ/pJM0029	This work
MD2062	rcsB70::Tn10d Cm spvA103::MudJ/pJM0029	This work
Plasmids
pIZ1589	pBAD18 rcsB+	27
pJM0029	pBAD18 rcsB(D56Q)	This work

Generation of a nonphosphorylatable RcsB(D56Q) variant.

A nonphosphorylatable RcsB(D56Q) variant was constructed using as the template the plasmid pIZ1589, a gift of J. Casadesús (Universidad de Sevilla, Spain). pIZ1589 expresses the rcsB⁺ wild-type allele under the control of the arabinose-inducible P_BAD promoter (27). Mutagenesis of the phosphorylatable aspartate residue at position 56 of RcsB was designed to replace codon 56, GAC, with codon CAG, specific for glutamine. The mutagenesis was performed by overlap extension PCR as described previously (32) using the following oligonucleotides: rcsB5′ (5′-AGC GGA ATT CAG GAG GAA TAC ATG AAC AAT ATG AAC G-3′), rcsB3′D56Q (5′-CAT GGA GAG CTG AGT GAT CAA C-3′), rcsB5′D56Q (5′-GTT GAT CAC TCA GCT CTC CAT G-3′), and rcsB3′ (5′-GTG AAA GCT TGT CGA CAA GCG ATT TAT TCT TTG TCT G-3′). The underlined oligonucleotide sequences are the sites changed in the mutagenesis procedure.

Construction of a ΔrcsD nonpolar deletion mutant.

To avoid any polar effect on rcsB transcription, the deletion of rcsD was designed from 155 nucleotides upstream of the start codon (micF-rcsD intergenic region) to position 2,523 of the 2,670-nucleotide open reading frame of rcsD. The deletion was created by following the one-step inactivation procedure described by Datsenko and Wanner and using plasmid pKD3 as the template (16). The oligonucleotides used for this procedure were rcsD-P1 (5′-AGC AAA TAA TTT CTT GAT ATT TAG TGC TAA ACA TTT ATA AGT AGT CTT TAT GTG TAG GCT GGA GCT GCT TC-3′) and rcsD-P2 (5′-TAA CTG CTT GCC GGG TAC CAG ATT AAG CAT GGC AAA CAC CCC TTT CAG GCG CAT ATG AAT ATC CTC CTT AGT-3′). Underlined in these sequences are the regions that anneal to the flanking regions of the cat cassette (Cm^r) present in plasmid pKD3. This allelic exchange generated the strain MD2022 (ΔrcsD69::cat). The cat cassette was further eliminated with plasmid pCP20, expressing the FLP recombinase, as described previously (16), and resulting in strain MD2023 (ΔrcsD69). Control experiments showed that rcsB transcription is not affected in strain MD2023 (see Fig. S1 in the supplemental material). Oligonucleotides specific for rcsD, rcsB, and rcsC used in these reverse transcription-PCR (RT-PCR) assays are listed in Table S1 in the supplemental material. The loss of the wild-type rcsD⁺ allele was confirmed with oligonucleotides yojN-4D and yojN-2R (see Table S1 in the supplemental material).

Phage transductions.

The transductional crosses were made using the P22 HT 105/1 int201 phage (51), as described previously (41). Phage-free transductants were screened on green plates as described previously (10) and tested for sensitivity to the clear-plaque mutant P22 H5.

Microarray analyses.

A 70-mer oligonucleotide microarray was constructed using the genome sequence of S. enterica serovar Typhimurium strain SL1344, made available by the Wellcome Trust Sanger Institute (see the supplemental material). Hybridization of nucleic acids was conducted using cDNA as the “expression sample” of the respective strain mixed with genomic DNA (gDNA) as the internal reference. As previously reported (21), this method allows transcriptome comparisons among different strains. RNA was purified in six independent experiments from bacteria grown in 10 ml of ISM-glycerol medium at exponential phase (optical density at 600 nm [OD₆₀₀], ∼0.2) by using the SV total RNA isolation system kit (catalog no. Z3100; Promega). These RNAs were combined in pools of three samples and used in two independent hybridizations. Prior to this manipulation, a 1:5 volume of a 95% ethanol-5% phenol solution was added to the bacterial culture (final concentration, 19% ethanol-1% phenol). As described previously, this step ensures the full integrity and stability of the RNA (21). The quality of the RNA was tested using an Agilent 2100 bioanalyzer (Agilent Technologies, Palo Alto, CA). RT of RNA was performed with SuperScript III reverse transcriptase (catalog no. 18080-044; Invitrogen) according to the instructions of the manufacturer. cDNA was purified with the QIAquick‰ PCR purification kit (catalog no. 28106; Qiagen). Reference gDNA was purified from bacteria grown in 6 ml of LB broth to stationary phase. Upon centrifugation (10,000 × g for 10 min, 4°C), the bacterial pellet was processed using the Qiagen gDNA kit (catalog no. 19060/10243; Qiagen). DNA was finally suspended in 200 μl of deionized water, resulting in a total amount in the range of 60 to 90 μg. This DNA was further fragmented by sonication before its use in the hybridization assays. Labeling of cDNA and gDNA with Alexa Fluor 647 and Alexa Fluor 555 fluorescence molecules, respectively, was conducted as described previously (21) using exo-Klenow enzyme and the BioPrime Plus array CGH indirect genomic labeling system (catalog no. 18096-011; Invitrogen). Labeled cDNA and gDNA were further purified, and the amount of incorporated labeling was estimated in a NanoDrop ND-1000 UV-visible-light spectrophotometer. Hybridization conditions, data acquisition, normalization, and statistical analysis are described in detail in the supplemental material.

RT-PCR assays.

RT-PCR was performed using a one-step RT-PCR kit (Qiagen) as previously described (55). Briefly, DNase (Ambion)-treated RNA samples were diluted to 20 ng/μl and were further normalized against the RT-PCR-amplified ompA product, used as an internal control. The RT-PCR was carried out with a final volume of 25 μl consisting of 5 μl buffer (5×), 1 μl of deoxynucleoside triphosphates (10 mM), 3 μl each of the forward and reverse primers (5 μM), 1 μl of RT-PCR enzyme mix, and ∼20 ng of RNA and RNase-free water supplemented to 25 μl. The RT-PCR cycling conditions were as follows: 50°C for 35 min and 95°C for 15 min, followed by 26 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 40 s and then an extra step of elongation at 72°C for 10 min. Oligonucleotides used in these RT-PCR assays are listed in Table S1 in the supplemental material.

β-Galactosidase assays.

The levels of β-galactosidase enzyme derived from the wcaH21::MudJ and spvA103::MudJ transcriptional fusions were determined using the CHCl₃-sodium dodecyl sulfate permeabilization procedure, as described previously (44). Bacteria were grown overnight at 37°C in LB broth or ISM-glycerol medium under shaking conditions (180 rpm), until a final OD₆₀₀ of ∼3.0 (LB medium) or ∼1.2 to 1.4 (ISM) was reached. Cultures were further diluted 1:100 in fresh medium and further incubated for 6 h, the time at which β-galactosidase activity was measured. In the case of RcsB variants expressed from plasmids under the arabinose-inducible P_BAD promoter, bacteria were grown overnight in ISM-2% glucose (OD₆₀₀, ∼1.2 to 1.4) and later washed once in ISM with no sugar added. These bacteria were then added to either ISM-2% glucose or ISM-glycerol-0.05% arabinose medium (final OD₆₀₀, ∼0.05) and further incubated for 16 h, the time at which the β-galactoside activity was measured. The experiments were repeated at least three times.

Virulence assays.

To assess the virulence capacity of IgaA-deficient strains with distinct levels of expression of the RcsCDB system, the protocol of Beuzón and Holden involving a 1:1 mixture of mutant and wild-type strains was used (2). Three female BALB/c mice (6 to 8 weeks old) were challenged per mixture using the intraperitoneal route. The experiment was repeated twice using 2 × 10⁵ CFU of each strain. The competitive index assay results were calculated from the numbers of viable bacteria counted in liver and spleen at 48 h postinfection.

Statistical analysis.

Data were analyzed by one-way analysis of variance using Prism version 5.0 (GraphPad Software, Inc.). Differences in the values with a P of <0.05 were considered statistically significant.

Accession numbers.

The characteristics and configuration of the Salgenomics microarray were deposited in the MIAME database (http://www.ebi.ac.uk/miamexpress) under accession number A-MEXP-846. Gene expression data were deposited in the Array Express database (http://www.ebi.ac.uk/arrayexpress) under accession number E-MEXP-1612.

RESULTS

Definition of the serovar Typhimurium Rcs regulon in a high-osmolarity minimal medium.

The Rcs regulon in E. coli has been extensively studied (23, 30, 46). In the case of serovar Typhimurium, a recent study reported microarray data of wild-type, rcsC, rcsD, rcsB, and rcsA strains exposed to sublethal concentrations of polymyxin B (20). The report of that study listed 26 serovar Typhimurium genes regulated by the RcsCDB system in an RcsA-independent manner (21 positively and 5 negatively). However, neither the identities of RcsB- and RcsA-coregulated genes nor the transcriptome profiles were provided. In another recent study, approximately 20% of the serovar Typhimurium genome was claimed to be under the control of RcsB in bacteria growing in nutrient-rich medium (55). In that study, a strain expressing an IgaA(T191P) mutant protein was used as a control of activation of the RcsCDB system.

In our attempt to dissect the modulation exerted by IgaA on the RcsCDB system, we performed genome expression profiling of serovar Typhimurium strains with defects in this attenuator and grown under conditions not tested in previous studies. We examined the effect of the R188H change in IgaA (allele igaA1), the first-characterized mutation that revealed a functional link between IgaA and the RcsCDB system (6, 7, 17). Bacteria were grown in the high-osmolarity minimal medium ISM (58), in which the Rcs response is more pronounced when IgaA is altered. Thus, the igaA1 mutant expressed the colanic capsule gene wcaH, positively regulated by the RcsCDB system, at ∼15-fold higher levels in ISM than in LB medium (Fig. 1). Our microarray study also included “mucoid” strains lacking IgaA which have recently been characterized (42). Two of these IgaA-defective strains with a functional RcsCDB system are MD0842-J4 (igaA2::KIXX rcsC67′) and MD0855-J10 [igaA2::KIXX Δ(ompC′-micF)], which spontaneously suppressed IgaA essentiality by diminishing the protein levels of RcsC and RcsD, respectively (42). In total, the isogenic strains used in our microarray studies included the wild type; igaA1, rcsB, rcsA, igaA1 rcsB, and igaA1 rcsA strains; MD0842-J4; and MD0855-J10 (Table 2). The microarray Salgenomics, designed on the genome sequence of the virulent serovar Typhimurium strain SL1344 (33), was constructed for this purpose (see the supplemental material). Only those gene expression changes with values of fourfold or higher (log₂ ratio, ≤−2 or ≥2) were considered significant.

FIG. 1.

Growth of serovar Typhimurium under high-osmolarity conditions augments the Rcs response resulting from defects in the attenuator IgaA. The expression of the colanic acid gene wcaH (gmm), positively regulated by the Rcs system, was monitored in bacteria grown in nutrient-rich LB medium (dark-gray bars) or ISM-glycerol minimal medium of high osmolarity (light-gray bars). The isogenic strains used were wild-type SV4407 (wcaH21::MudJ), MD0099 (igaA1 wcaH21::MudJ), and MD0870 (igaA1 rcsB70 wcaH21::MudJ). These strains are abbreviated in the figure as WT, igaA, and igaA1 rcsB, respectively. Bacteria were grown to an OD₆₀₀ of ∼2.5 to ∼3.0 (LB medium) or ∼1.2 to ∼1.4 (ISM-glycerol medium). β-Galactosidase activities correspond to the means and standard deviations from a minimum of three independent experiments. Mean values are also shown on top of the bars.

TABLE 2.

List of the tables in the supplemental material showing serovar Typhimurium genes with altered expression caused by mutations in igaA, components of the RcsCDB system, and/or the coregulator RcsA^a

Table in the supplemental material	Strains used for transcriptome comparisonb
S2	All strains
S3	igaA1 mutant vs igaA1 rcsB mutant
S4	rscB mutant vs wild type
S5	igaA1 rcsB mutant vs wild type
S6	igaA1 mutant vs igaA1 rcsA mutant
S7	igaA1 rcsA mutant vs wild type
S8	rcsA mutant vs wild type
S9	igaA2::KIXX rcsC′ mutant vs wild type
S10	igaA2::KIXX (ΔompC′-micF) mutant vs wild type
S11	igaA1 mutant vs igaA2::KIXX rcsC′ mutant
S12	igaA1 mutant vs igaA2::KIXX (ΔompC-micF) mutant
S13	igaA2::KIXX (ΔompC-micF) mutant vs igaA2::KIXX rcsC′ mutant

^aAltered expression is indicated by a P value of ≤0.05 and a log₂ ratio that was less than or equal to −2 or ≥2.

^bStrain designations for the indicated genotypes are SL1344 (wild type), SV4450 (igaA1 mutant), SV4379 (rcsA mutant), SV4406 (rcsB mutant), SV4343 (igaA1 rcsA mutant), MD0862 (igaA1 rcsB mutant), MD0842-J4 (igaA2::KIXX rcsC′ mutant), and MD0855-J10 [igaA2::KIXX (ΔompC′ -micF) mutant]. See also Table 1. Genes with differential expression in the comparison of the igaA1 mutant and the wild type are shown in Tables 3 and 4.

Identification of IgaA-regulated genes.

The effect of the IgaA(R188H) mutation in the Rcs regulon was assessed in the wild-type, igaA1, rcsB, and igaA1 rcsB strains. Expression profiles were processed in the form of comparisons of a mutant and the wild type or of two mutants (Table 2). A total of 71 genes displaying changes in expression from the wild-type strain of at least fourfold were identified in the igaA1 mutant. Of these, 33 were upregulated (Table 3) and 38 downregulated (Table 4). Most of the upregulated genes (n = 22), are involved in colanic acid capsule synthesis (wza-wzb, wcaABCDEF-gmd-wcaGHI-manC, cpsG-wcaJ, wzxC-wcaKLM, and the regulatory gene rcsA) (30, 55). The yjbEFGH operon, recently described to be required for the synthesis of a novel exopolysaccharide in E. coli (22), was also upregulated in the serovar Typhimurium igaA1 mutant (Table 3). So, ∼80% (26/33) of the genes highly upregulated in the igaA1 mutant are involved in the synthesis of capsular exopolysaccharides. Other genes induced in the igaA1 mutant and previously known to be regulated by RcsCDB are osmB, ygaC, and yebE (3, 23, 30, 55). On the other hand, metB-metF, two genes involved in methionine biosynthesis, were identified in our study as new genes positively regulated by the RcsCDB system.

TABLE 3.

Genes in the serovar Typhimurium igaA1 mutant with increased expression relative to that in the wild-type strain

Systematic gene name	Gene	Product	Log2 ratio for the indicated comparisona					Reference(s) if the product was previously assigned to the Rcs regulonb
			igaA1 mutant vs WT	igaA1 rcsB mutant vs WT	rcsB mutant vs WT	igaA1 mutant vs igaA1 rcsA mutant	igaA1 rcsA mutant vs WT
STM1633	STM1633	Putative periplasmic binding protein	2.3	−1.9	−2.2	0.1	2.2	55
STM1705	osmB	Lipoprotein B	3.0	−0.9	−0.05	−1.0	4.1	23, 30, 55
STM1880	yebE	Putative inner membrane protein	2.2	−0.06	0.1	1.3	0.9
STM1982	rcsA	Colanic acid capsular regulator A	3.6	−1.6	−1.2	2.1	1.7	23, 55
STM1983	dsrB	Hypothetical protein STM1983	2.4	−0.7	−0.9	1.2	1.2
STM2080	udg	UDP-glucose/GDP-mannose dehydrogenase	2.5	−0.8	−1.1	2.5	−0.2	23, 55
STM2099	wcaM	Putative colanic acid biosynthetic protein	3.3	−0.6	−1.0	3.0	0.5	23, 30, 55
STM2100	wcaL	Putative glycosyl transferase	6.4	0.1	0.9	2.0	4.4	30, 55
STM2101	wcaK	Putative galactokinase	6.6	−0.6	−0.1	3.2	3.3	30, 55
STM2102	wzxC	Putative colanic acid exporter	5.4	−0.1	−0.2	2.9	2.7	23, 55
STM2103	wcaJ	UDP-glucose lipid carrier transferase	5.6	−0.9	−0.2	2.9	2.8	30, 55
STM2104	cpsG	Phosphomannomutase	7.0	−0.9	0.5	3.2	3.8	30, 55
STM2105.S	manC	Mannose-1-phosphate guanylyl transferase	8.0	−0.5	0.5	3.1	4.6	23, 55
STM2106	wcaI	Putative glycosyl transferase	8.1	−0.1	1.0	3.2	4.7	23, 30, 55
STM2107	wcaH	GDP-mannose mannosyl hydrolase	8.3	−0.8	1.0	3.6	4.5	23, 30, 55
STM2108	wcaG	GDP-fucose synthetase	8.0	−0.6	−0.4	3.5	4.4	23, 30, 55
STM2109	gmd	GDP-D-mannose dehydratase	8.6	−1.7	−1.2	3.7	4.3	23, 30, 55
STM2110	wcaF	Putative acyl transferase	7.6	−0.7	−0.9	4.0	3.6	23, 30, 55
STM2111	wcaE	Putative transferase	6.9	−0.7	0.1	4.1	2.7	23, 30, 55
STM2112	wcaD	Putative colanic acid polymerase	6.5	−0.3	−0.5	4.7	2.0	23, 30, 55
STM2113	wcaC	Putative glycosyl transferase	4.6	−0.6	−0.1	4.1	0.8	23, 30, 55
STM2114	wcaB	Putative acyl transferase	5.5	−0.2	−0.2	3.6	1.9	30, 55
STM2115	wcaA	Putative glycosyl transferase	4.3	0.03	2.1	3.9	0.4	23, 30, 55
STM2117	wzb	Putative phosphotyrosine protein phosphatase	5.8	−0.9	−0.4	5.0	0.9	23, 55
STM2118	wza	Putative outer membrane polysaccharide export protein	5.1	−0.7	−0.7	5.1	0.2	23, 55
STM2407	ypeC	Putative periplasmic protein	2.2	−0.7	−0.6	−0.02	2.4
STM2801	ygaC	Putative cytoplasmic protein	3.0	−1.5	−2.3	0.4	2.9	23, 55
STM4100	metB	Cystathionine gamma synthase	2.4	0.2	0.5	−0.4	3.0
STM4105	metF	5,10-Methylenetetrahydrofolate reductase	2.5	0.5	0.2	0.4	2.1
STM4222.S	yjbE	Putative outer membrane protein	5.9	−0.2	−0.1	4.3	1.7	20, 22, 23, 55
STM4223	yjbF	Putative outer membrane lipoprotein	5.9	−0.5	0.3	3.2	2.6	22, 23, 55
STM4224	yjbG	Putative periplasmic protein	5.0	−0.7	−0.9	3.6	1.6	22, 23, 55
STM4225	yjbH	Putative outer membrane lipoprotein	3.2	−0.2	−0.9	2.4	1.0	22, 23, 27, 55

^aEach gene was represented by a single 70-mer oligonucleotide spotted in two separate sections of a microarray slide (see the supplemental material for details). For the statistical analyses, the two spots were considered different entities. The table lists only those genes identified in the comparison of the igaA1 mutant and the wild type (WT), both of whose P values were ≤0.05 and whose two log₂ ratios were ≥2. For simplification of this table, only the average of the two log₂ ratios for each gene is shown (see Table S2 in the supplemental material for a complete list of the log₂ ratios and P values).

^bNo reference indicates no previous assignment.

TABLE 4.

Genes with decreased expression in the serovar Typhimurium igaA1 mutant versus the wild-type strain

Systematic gene name	Gene	Product	Log2 ratio for the indicated comparisona					Reverence(s) if the product was previously assigned to the Rcs regulon
			igaA1 mutant vs WT	igaA1 rcsB mutant vs WT	rcsB mutant vs WT	igaA1 mutant vs igaA1 rcsA mutant	igaA1 rcsA mutant vs WT
PSLT039	spvB	Hydrophilic protein	−4.0	−4.5	−4.2	−0.9	−3.1
PSLT040	spvA	Outer membrane protein	−5.1	−5.2	−4.6	−1.8	−3.2
STM0462	glnK	Nitrogen regulatory protein P-II 2	−2.6	0.6	0.7	−2.8	−0.04
STM0829	glnP	Glutamine ABC transporter permease component	−2.5	0.3	0.1	−1.7	−0.9
STM1172	flgM	Antiflia factor	−2.9	−1.6	−1.2	0.4	−3.5	55
STM1174	flgB	Flagellar basal body rod protein	−3.0	−0.9	−1.0	1.5	−5.0	20, 55
STM1175	flgC	Flagellar basal body rod protein	−3.9	−2.1	−2.0	1.6	−5.8	46, 55
STM1176	flgD	Flagellar basal body rod modification protein	−3.6	−1.9	−1.8	1.6	−5.5	20, 55
STM1177	flgE	Flagellar hook protein	−3.5	−1.8	−1.4	1.2	−5.1	55
STM1178	flgF	Cell-proximal portion of basal body rod	−3.1	−1.7	−1.7	1.8	−5.3	55
STM1179	flgG	Flagellar basal body rod protein	−3.4	−1.5	−1.5	0.4	−3.9	46, 55
STM1181	flgI	Flagellar P-ring protein precursor	−2.5	−1.3	−1.7	1.3	−4.0	46, 55
STM1182	flgJ	Flagellar biosynthesis protein	−2.5	−1.2	−0.6	0.2	−2.8	55
STM1183	flgK	Flagellar hook-associated protein	−3.1	−1.5	−1.5	1.1	−4.4	55
STM1916	cheY	Chemotaxis regulator	−3.0	−1.5	−1.5	0.8	−4.0	55
STM1917	cheB	Chemotaxis-specific methylesterase	−2.5	−1.5	−1.2	1.1	−3.9	55
STM1919	cheM	Methyl-accepting chemotaxis protein II	−3.1	−1.4	−1.4	−0.06	−3.2	55
STM1921	cheA	Chemotaxis sensory histidine protein kinase	−2.7	−1.0	−1.4	−0.07	−2.8	55
STM1922	motB	Flagellar motor protein	−2.2	−1.2	−0.9	1.4	−3.8	55
STM1956	fliA	Flagellar biosynthesis sigma factor flia	−2.7	−1.7	−1.2	1.2	−4.1	55
STM1959	fliC	Flagellar biosynthesis protein	−3.1	−1.3	−1.5	1.9	−5.3	55
STM1960	fliD	Flagellar hook-associated protein	−3.2	−1.2	−0.8	0.3	−3.8	55
STM1961	fliS	Flagellar protein flis	−2.2	−0.8	−0.9	0.7	−3.2	55
STM1962	fliT	Possible flid export chaperone	−2.3	−0.7	−1.1	0.8	−3.3	55
STM1970	fliG	Flagellar motor protein	−2.2	−1.5	−0.8	0.06	−2.5	55
STM1971	fliH	Flagellar assembly protein	−2.5	−1.2	−0.9	0.09	−2.7	55
STM1973	fliJ	Flagellar protein	−2.3	−1.3	−1.4	0.2	−2.6	55
STM1974	fliK	Flagellar hook length control protein	−2.1	−1.4	−1.2	0.1	−2.4	55
STM1975	fliL	Flagellar biosynthesis protein	−2.3	−1.3	−1.4	0.1	−2.5	55
STM2259	napA	Periplasmic nitrate reductase	−3.7	0.7	0.8	−0.4	−3.3
STM2260	napD	Periplasmic nitrate reductase	−2.7	0.6	0.2	0.1	−2.9
STM2283	glpT	Sn-glycerol-3-phosphate transport protein	−2.2	−0.05	−0.2	0.2	−2.3	23
STM2284	glpA	Sn-glycerol-3-phosphate dehydrogenase large subunit	−2.8	0.7	0.7	0.2	−3.1
STM2771	fljB	Flagellar biosynthesis protein	−3.1	−1.3	−1.5	0.9	−2.0	55
STM3274	yhbU	Putative protease	−2.3	0.8	0.2	−0.6	−1.5
STM3474	nirB	Nitrite reductase large subunit	−2.5	0.6	0.2	0.4	−3.1
STM4229	malE	Periplasmic maltose-binding protein	−2.4	−1.7	−1.4	−0.8	−1.7	46
STM4231	lamB	Maltoporin precursor	−2.7	−2.1	−1.6	−0.5	−2.3	46

^aEach gene was represented by a single 70-mer oligonucleotide spotted in two separate sections of a microarray slide (see the supplemental material for details). For the statistical analyses, the two spots were considered different entities. The table lists only those genes identified in the comparison of the igaA1 mutant and the wild type, both of whose P values were ≤0.05 and whose two log₂ ratios were ≤2. For simplification of this table, only the average of the two log₂ ratios for each gene is shown (see Table S2 in the supplemental material for a complete list of log₂ ratios and P values).

^bNo reference indicates no previous assignment.

Among the 38 downregulated genes in the igaA1 mutant, 26 were flagellum and chemotaxis genes (Table 4). Unlike in a recent study that analyzed the response in LB medium of an igaA(T191P) mutant with high Rcs activity (55), no reduced expression of virulence genes present in Salmonella pathogenicity islands 1 and 2 or of fimbrial genes was noted. Interestingly, other genes repressed in our igaA1 mutant are related to ammonia metabolism and nitrate/nitrite respiration (glnK, glnP, napA, napD, nirB), transport of maltose (malE, lamB), and uptake/metabolism of glycerol-3-phosphate (glpT, glpA). In addition, a pronounced repression of spvA and spvB, two Salmonella-specific genes required for virulence (29), was detected in the igaA1 mutant with log₂ ratios in the range of −4 or −5, equivalent to ∼15- to 30-fold changes (Table 4). A search in the literature for genes previously assigned to Rcs regulons of different enterobacteria revealed that among the 71 genes identified with altered expression in the igaA1 mutant, 13 were not reported before as genes regulated by the RcsCBD system. These include yebE, dsrB, ypeC, metB, metF, spvB, spvA, glnK, glnP, napA, napD, glpA, yhbU, and nirB (Tables 3 and 4).

The gene expression profile of the igaA1 mutant was then compared to those of the igaA1 rcsB and rcsB strains. Among the 71 genes having altered expression in the igaA1 mutant compared to that in wild-type bacteria (Tables 3 and 4), 68 of them did not display any significant change in expression in the igaA1 rcsB and rcsB strains (Fig. 2A and B). It is noteworthy that the yciGFE operon and the Salmonella-specific genes STM1862 (pagO), STM1863, and STM1864 were downregulated only in the mutants lacking RcsB, irrespective of the functional status of IgaA (igaA1 rcsB and rcsB mutants) (Fig. 2B; see also Tables S3, S4, and S5 in the supplemental material). Such observations may indicate that these genes are maximally expressed with low levels of RscB∼P. Other genes, such as spvA and spvB, were, however, downregulated to a large extent in the three strains tested, namely, the igaA1, igaA1 rcsB, and rcsB mutants (Table 4; Fig. 2B). To further confirm these observations, RT-PCR was used as an alternative method to monitor the expression of representative genes not described before as regulated by the RcsCDB system (spvA, metB, ypeC, glnK, glnP, napA, glpA, and nirB). Control genes known to form part of the Rcs regulon, such as wcaH and ygaC, were also included. The results of these RT-PCR assays were in agreement with the microarray data, with expression profiles for the igaA1 mutant being clearly opposite to those of the wild-type, igaA1 rcsB, and rcsB strains, except in the expression of spvA (Fig. 3). Taken together, these studies demonstrated that in bacteria growing at high osmolarity, IgaA antagonizes the activity of RcsB∼P for most genes that are targeted by this response regulator.

FIG. 2.

The attenuator IgaA controls part of the Rcs regulon. The Venn diagrams show genes induced (A) or repressed (B) in igaA1, igaA1 rcsB, and rcsB strains upon comparison of their respective gene expression profiles with that of the wild-type strain. Only genes displaying changes with a log₂ ratio that was ≤−2 or ≥2 are shown. Some genes, such as STM1863-STM1864 or yciGFE, are downregulated when RcsB is not functional but not induced in the igaA1 mutant with a strong RcsCDB response. Other genes, such as spvAB, display a complex regulatory pattern since they are positively regulated by RcsB (repressed in the igaA rcsB and rcsB strains) but also downregulated when the RcsC→RcsD→RcsB phosphorelay is stimulated (igaA1 strain). The asterisk after rcsB indicates that the gene is knocked out in the two mutants (igaA1 rcsB and rcsB mutants). Genes labeled with two asterisks indicate cases in which, for some of the comparisons shown, one of the log₂ ratios of the gene duplicates was less than or equal to −2 while the other value was close to −2. Thus, STM1862 (pagO) and phoN could be assigned to the category of genes repressed by RcsB: their log₂ ratios in the comparison of the rcsB mutant with the wild type were −2.96/−1.69 and −2.4/−1.8, respectively. Likewise, spvC and spvR could be assigned to the category of genes repressed in the igaA1, igaA1 rcsB, and rcsB strains. spvC had log₂ ratios of −2.8/-1.1 and −2.7/-1.5 in the comparisons of the igaA1 mutant with the wild type and of the rcsB mutant with the wild type, respectively. spvR had log₂ ratios of −2.2/−1.4 in the comparison of the igaA1 mutant with the wild type (see Table S2 in the supplemental material for a complete list of genes).

FIG. 3.

Altered expression in the igaA1 mutant of genes not previously assigned to the RcsCDB regulatory system. Shown are the results of RT-PCR assays of representative genes that the microarray experiments identified as up- or downregulated in the igaA1 mutant (Tables 3 and 4). wcaH, required for synthesis of the colanic acid capsule, and ygaC were used as controls for genes positively regulated by the RcsCDB system. Note that, in concordance with the microarray data, spvA is downregulated in the igaA1 mutant and in strains lacking RcsB. RNA was isolated from bacteria grown in ISM-glycerol medium at an OD₆₀₀ of ∼1.2 to ∼1.4. Numbers indicate the average log₂ ratios shown in Tables 3 and 4 for each of the indicated strains (igaA1, igaA1 rcsB, and rcsB strains) and relative to values for the wild type (WT).

Identification of genes coregulated by RcsA and RcsB.

RcsA coregulates with RcsB the expression of colanic acid capsule genes and a distinct exopolysaccharide synthesized by the products of the yjbEFGH operon (22, 39, 55). To define the effect of igaA mutations on the contribution of RcsA in serovar Typhimurium, we monitored transcription profiles in igaA1, igaA1 rcsA, and rcsA strains. The genes displaying altered transcription were differentiated into two classes on the basis of the strict requirement of the coregulator RcsA for expression. The first class contained genes that were entirely dependent on RcsA, i.e., with altered expression in igaA1 rcsA and igaA1 strains. This class included the positively regulated colanic acid genes wcaM, wza-wzb, and wcaA-wcaB-wcaC-wcaD; the yjbEFGH operon; and yebE (Table 3; see also Table S6 in the supplemental material). Within this category, but negatively regulated, were glnK and glnP (Table 4; see also Table S6 in the supplemental material). To our knowledge, this result represents the first evidence of RcsA being involved in negatively regulating metabolic genes. In a second category, we classified the genes in the transcriptome of the igaA1 rcsA strain with altered expression levels compared with those in the wild type. Within this class, we identified upregulated genes, such as the operons wcaE-wcaF-gmd-wcaG-wcaH-wcaI and cpsG-wcaJ-wzxC-wcaK-wcaL, involved in colanic acid synthesis (Table 3; see also Table S7 in the supplemental material). This result suggests that RcsB∼P alone, probably at high levels as a consequence of the igaA1 mutation, can drive the expression of a certain proportion of colanic acid capsule genes in an RcsA-independent manner. Other conditions leading to enhanced RcsB∼P levels, such as the expression of the constitutive mutant allele rcsC137 or the expression of rcsB⁺ in multicopy, have previously been shown to render the expression of colanic genes independent of RcsA (4). A detailed view of our expression profiles revealed, however, that this set of colanic capsule genes (wcaE-wcaF-gmd-wcaG-wcaH-wcaI and cpsG-wcaJ-wzxC-wcaK-wcaL) are expressed in the igaA1 mutant to a higher extent than in the igaA1 rcsA strain (Table 3; see also Table S6 in the supplemental material). Lastly, note that some of the RcsB-regulated genes expressed in an IgaA⁺ (wild-type) background, such as yciGFE (Fig. 2; see also Table S4 in the supplemental material), also require RcsA for expression (see Table S8 in the supplemental material). All together, these data indicate that RcsA controls part of the Rcs regulon of serovar Typhimurium and that genes belonging to this class can be further subdivided by their capacity to be expressed by RcsB∼P in the absence of this coregulator.

A decreased amount of RcsC or RcsD results in different Rcs responses.

We have recently shown that serovar Typhimurium can spontaneously suppress the requirement of the essential protein IgaA by acquiring deletions affecting the integrity of the rcsCDB locus (42). Two of the characterized suppressor strains lacking IgaA, MD0842-J4 and MD0855-J10, retain activity in the RcsCDB system since both display increased expression of the colanic acid gene wcaH, produce altered levels of the flagellins FliC and FljB and the cell division proteins FtsA and FtsZ, and are mucoid on plates (42). As a result of deletions affecting the regulatory regions of rcsC and rcsD, MD0842-J4 and MD0855-J10 mutants produce small amounts of RcsC and RcsD, respectively (42). To obtain insights into how the RcsCDB system could operate in the absence of IgaA, we monitored genome expression profiles in MD0842-J4 and MD0855-J10. The two mutants showed changes similar to those observed in the igaA1 mutant and indicative of the activity of the RcsCDB system, including the upregulation of colanic capsule genes and repression of flagellar genes (see Tables S9 to S13 in the supplemental material). Nonetheless, several new features differentiated their Rcs responses. The IgaA-defective strain MD0842-J4, with low levels of RcsC, showed reduced expression of colanic capsule genes compared to that of the igaA1 mutant (see Table S11 in the supplemental material). This result was consistent with the higher expression of a wcaH::lacZ transcriptional fusion in the igaA1 mutant than in MD0842-J4 (42). Unlike the igaA1 mutant, the MD0842-J4 mutant, however, expressed spvAB at wild-type levels (see Table S9 in the supplemental material). In contrast, the IgaA-defective strain MD0855-J10, which has low levels of RcsD, displayed reduced spvAB expression (see Table S10 in the supplemental material). As expected by the pronounced mucoidy displayed by MD0855-J10 on plates, the expression of the colanic capsule genes in this mutant was comparable to that observed in the igaA1 mutant (see Table S12 in the supplemental material). Globally, these data indicate that a decrease in RcsC significantly reduces the responsiveness of the RcsCDB system, even in the absence of the attenuator IgaA. This effect is evident in the case of colanic acid genes, which are induced only at intermediate levels, and the spvAB genes, whose expression remains unaltered. Conversely, the RcsCDB system can reach high activation levels with decreased amounts of RcsD, leading to high expression of colanic acid genes and repression of the spvAB genes.

A decreased level of RcsC or RcsD in IgaA-deficient strains has different impacts on serovar Typhimurium virulence.

An inverse correlation between RcsCDB activity and the capacity of serovar Typhimurium to cause disease in the mouse acute-infection model has previously been shown (17, 45). The microarray data obtained with the IgaA-deficient strains MD0842-J4 and MD0855-J10 revealed differences in spvAB expression, and as previously mentioned, these two strains display Rcs responses of different intensities (42). Since both MD0842-J4 and MD0855-J10 are mucoid, it was of interest to test whether these strains behave differently in the mouse typhoid virulence model. Intraperitoneal-challenge assays showed that the MD0855-J10 strain, with a higher Rcs response (42), was highly attenuated but that the MD0842-J4 strain, with lower Rcs activity, was virulent (Table 5). We included the MD0835-J1 mutant as a control strain, which also lacks IgaA but harbors a deletion of the apbE′-ompC-micF-rcsD-rcsB-rscC′ region (42). This MD0835-J1 mutant behaved as a virulent strain (Table 5); therefore, we discarded the possibility that the attenuation of the MD0855-J10 mutant, which harbors an ompC-micF-rcsD′ deletion (42), was due to the lack of ompC and/or micF. These data reinforce the idea of IgaA modulating virulence exclusively at the stage of controlling the responsiveness of the RcsCDB system. In IgaA-defective bacteria, virulence can differ significantly depending on whether the relative amount of either RcsC or RcsD is compromised.

TABLE 5.

Virulence of IgaA-deficient mutants of serovar Typhimurium having distinct activation levels of the RcsCDB system

Strain mixture	Status of the Rcs systema	Competitive indexb
		Liver	Spleen
MD0835-J1 [igaA2::KIXX Δ(apbE′-rcsC′)] × WT	Inactive	0.63 ± 0.11	0.82 ± 0.12
MD0842-J4 [igaA2::KIXX rcsC67′] × WT	Intermediate	0.65 ± 0.03	0.92 ± 0.11
MD0855-J10 [igaA2::KIXX Δ(ompC′-micF)] × WT	High	0.06 ± 0.02	0.07 ± 0.01

^aAs indicated by the relative levels of expression of the wcaH21::lacZ transcriptional fusion in the MD0835-J1, MD0842-J4, and MD0855-J10 strains (see reference 42).

^bCompetitive index values were calculated as described previously (2) after intraperitoneal challenge of BALB/c mice with a dose of 2 × 10⁵ CFU per strain. Liver and spleen were extracted at 48 h postinfection. Shown are the medians and standard deviations corresponding to results from two independent experiments.

spvA expression is positively controlled by RcsB but affected negatively by the RcsC→RcsD→RcsB phosphorelay.

The microarray data uncovered an unexpected complexity in the regulation of the virulence genes spvAB. These genes were downregulated in the igaA1 mutant upon activation of the RcsCDB system but were also repressed in bacteria lacking RcsB (Table 4; Fig. 2). Furthermore, spvAB displayed altered expression in IgaA-deficient bacteria but only in the mutant having reduced levels of RcsD, MD0855-J10 (see Table S10 in the supplemental material). Based on these observations, we hypothesized that spvAB could be subjected to positive regulation by dephosphorylated RcsB. In this context, any increase in the RcsB∼P/RcsB ratio could therefore lead to spvAB repression. This postulate predicts that normal spvAB expression levels should be restored in the igaA1 mutant upon inactivation of either RcsC or RcsD. To test this hypothesis, expression of an spvA::lacZ transcriptional fusion was monitored in the wild-type, igaA1, igaA1 rcsB, rcsB, igaA rcsC, rcsC, igaA1 rcsD, igaA1 rcsA, rcsA, rcsD, MD0842-J4, and MD0855-J10 strains. All of the mutants were characterized in previous studies (6, 17, 42), except those containing the rcsD null mutation, which we generated for this work. The rcsD deletion encompassed the entire rcsD open reading frame except the last 130 nucleotides at the 3′ end, which were left intact to preserve intact the 5′ regulatory region of rcsB. Control assays revealed that this deletion does not affect rcsB transcription (see Fig. S1 in the supplemental material). Figure 4 shows that spvA transcription decreases significantly when the level of the RcsC→RcsD→RcsB phosphorelay is high (igaA1, igaA1 rcsA, and MD0855-J10 strains). Interestingly, rcsC and rcsD null mutations, but not the rcsA mutation, restored wild-type levels of expression of spvA in the igaA1 background (Fig. 4). This result suggested that the repression of spvA observed in the igaA1 mutant requires an active RcsC→RcsD→RcsB phosphorelay but is independent of RcsA. Since the loss of RcsC or RcsD alone did not have any effect on spvA transcription, but RcsB was essential for such expression, we conclude that dephosphorylated RcsB may contribute to positively regulate this virulence gene.

FIG. 4.

Expression of the serovar Typhimurium virulence gene spvA requires RcsB but is impaired upon stimulation of the RcsCDB system. spvA expression was monitored in bacteria grown in ISM-38 mM glycerol (OD₆₀₀, ∼1.2 to ∼1.4). The following isogenic strains were used: SV4312 (spvA103::MudJ), MD1368 (igaA1 spvA103::MudJ), MD1369 (igaA1 rcsB70 spvA103::MudJ), MD1370 (rcsB70 spvA103::MudJ), MD2008 (igaA1 rcsC52 spvA103::MudJ), MD2007 (rcsC52 spvA103::MudJ), MD2028 (igaA1 ΔrcsD69 spvA103::MudJ), MD2027 (ΔrcsD69 spvA103::MudJ), MD1396 (igaA1 rcsA51 spvA103::MudJ), MD1397 (rcsA51 spvA103::MudJ), MD1371-MD0842-J4 (J4) (igaA3::Cam^r rcsC67′ spvA103::MudJ), and MD1372-D0855-J10 (J10) [igaA3::Cam^r Δ(ompC′-micF) spvA103::MudJ]. The genotype of each of these strains is abbreviated in the figure. The β-galactosidase activities correspond to the means and standard deviations from a minimum of three independent experiments and are shown as relative to the wild-type level (100%). The β-galactosidase activity of the wild-type strain (WT) had a mean value of 131.6 Miller units, as determined from a total of nine independent experiments. Note that spvA expression is repressed in strains with high activity of the RcsC→RcsD→RcsB phosphorelay (igaA1 mutant, igaA1 rcsA mutant, and MD0855-J10) and in those lacking RcsB (igaA1 rcsB and rcsB mutants).

Production of a nonphosphorylatable RcsB(D56Q) variant enhances spvA expression.

Data shown in Fig. 4 suggested that dephosphorylated RcsB could play a role in the positive regulation of spvA. However, other interpretations are also possible, such as that spvA was activated at low RcsB∼P levels but repressed at high levels of RcsB∼P or that RcsB was phosphorylated by other phospho-donors, such as acetyl-phosphate in the rcsC or rcsD mutant. To address this point, we used plasmids in which RcsB⁺ and a nonphosphorylatable RcsB(D56Q) variant were produced from the arabinose-inducible P_BAD promoter. To confirm the activities and correct expression of these two RcsB versions, they were first analyzed in an igaA rcsB strain. As expected, RcsB⁺ but not RcsB(D56Q) restored mucoidy in bacteria grown in medium supplemented with arabinose (Fig. 5A). RcsB⁺ or RcsB(D56Q) were then produced in wild-type, igaA1, igaA1 rcsB, and rcsB strains carrying the spvA::lacZ transcriptional fusion (Fig. 5B). When spvA expression was compared in media containing either glucose or arabinose, two phenomena were observed. Unlike with what was observed in bacteria grown with glycerol as the only carbon source (Fig. 4), low spvA expression was detected in all of the different strains grown in glucose-containing medium (see the legend to Fig. 5). This effect did not impair our analysis of the relative changes in spvA expression upon the production of RcsB from plasmid (arabinose versus glucose). The expression of the nonphosphorylatable RcsB(D56Q) variant, but not of RcsB⁺, drastically increased spvA expression (Fig. 5B). This effect, noticeable in strains lacking endogenous RcsB, was, however, more pronounced in strains with an rcsB⁺ background. Thus, when RcsB(D56Q) was produced in wild-type and igaA1 strains, spvA expression increased up to 22- and 90-fold, respectively (Fig. 5B). In silico analysis of the promoter regions of spvA and spvR, the latter encoding a positive regulator of the spvABC operon (59), did not reveal any sequence close to the consensus RcsAB box (56; data not shown). Taken together, these observations unequivocally demonstrated the capacity of dephosphorylated RcsB to stimulate spvA expression. In addition, they indicated that RcsB∼P may also be required to achieve proper spvA expression levels.

FIG. 5.

Expression of a nonphosphorylatable RcsB(D56Q) variant enhances spvA expression to a higher extent than wild-type RcsB⁺. These RcsB proteins were produced from plasmids under the arabinose-inducible P_BAD promoter (Table 1). (A) Control experiment showing that the expression of wild-type RcsB⁺ but not that of the nonphosphorylatable RcsB(D56Q) variant restores the activity of the RcsCDB system, as denoted by the mucoidy displayed by bacteria on plates. (B) Expression of spvA in wild-type, igaA1, igaA1 rcsB, and rcsB strains producing either RcsB⁺ or RcsB(D56Q). For every strain, the β-galactosidase activity measured under inducing conditions (16 h in ISM-0.05% arabinose-38 mM glycerol) is relative to that obtained in glucose-containing medium (16 h in ISM-2% glucose), which was normalized to 1. The β-galactosidase activity in Miller units measured for each strain in ISM-2% glucose was 7.4 (wild-type [WT] P_BAD-rcsB⁺ strain), 6.2 (igaA1 P_BAD-rcsB⁺ strain), 3.0 (igaA1 rcsB P_BAD-rcsB⁺ strain), 17.3 (rcsB P_BAD-rcsB⁺ strain), 9.6 [wild-type P_BAD-rcsB(D56Q) strain], 4.8 [igaA1 P_BAD-rcsB(D56Q) strain], 9.3 [igaA1 rcsB P_BAD-rcsB(D56Q) strain], and 9.2 [rcsB P_BAD-rcsB(D56Q) strain]. Data correspond to the means and standard deviations from three independent experiments.

DISCUSSION

The RcsCDB system is a complex regulatory system playing an important role in biofilm formation, virulence, and adaptation to envelope stress (34, 39). Genome expression studies, in fact, show that most of the genes of the Rcs regulon encode proteins related to envelope homeostasis (20, 23, 30, 31, 46, 55). Here, we also studied the role of IgaA as an attenuator of the RcsCDB system with a strategy based on genome expression profiling. On this occasion, we used a large number of serovar Typhimurium igaA mutants and exposed the bacteria to high osmolarity, a stimulus known to exacerbate the activity of the RcsCDB system. Under these particular growth conditions, 71 genes in the igaA1 mutant, 60 genes in the IgaA-defective MD0842-J4 strain (low RcsC levels), and 47 genes in the IgaA-defective MD0855-J10 mutant (low RcsD levels) displayed changes in expression higher than fourfold. The analysis of functions equally affected in the three mutants defines an Rcs regulon of at least 89 genes, approximately 2% of the serovar Typhimurium genome.

In qualitative terms, the Rcs regulon characterized in this work is similar to that reported in previous studies, with most genes encoding envelope and motility or chemotaxis functions (23, 30, 46). However, our study uncovered new regulatory features and identified up to 13 novel genes regulated by the RcsCDB system (Tables 3 and 4). Some of these new genes encode hypothetical proteins, while others are involved in processes related to glycerol-3-phosphate metabolism, methionine biosynthesis, ammonia metabolism, or nitrate/nitrite respiration. Some of these processes, such as those linked to the metabolism of ammonia, nitrate, and nitrite, are repressed upon activation of the RcsCDB system. These observations suggest that the role of the RcsCDB system might extend beyond the control of motility, chemotaxis, and envelope components to other scenarios related to metabolic readjustments in response to stress.

Our data also revealed that the RcsB∼P activity present in wild-type bacteria may be sufficient to account for maximal expression of genes such as STM1862 (pagO)-STM1863-STM1864 and yciGFE. Other cases in which higher expression is detected in the wild type than in the rcsB mutant are known for the gadAB genes in E. coli, required for glutamate-dependent acid resistance (9), and for srfA and siiE in serovar Typhimurium (27). More recently, STM1841, a serovar Typhimurium gene of unknown function, was also shown to have a 30-fold-higher induction in the wild type than in the rcsB mutant (55). These observations reflect that some genes of the Rcs regulon may require only low RcsB∼P levels to be maximally expressed but that the rest are gradually upregulated as the relative RcsB∼P levels increase in response to stimuli.

We were also interested in dissecting the contribution of RcsA to the Rcs regulon of serovar Typhimurium. rcsA was highly expressed in the igaA1 mutant, confirming the auto-regulation proposed to act via an RcsAB box present in the rcsA promoter (18, 56). Two classes of RcsA-regulated genes, one entirely dependent on this coregulator, were identified. However, we did not observe any correlation of these two categories with the magnitude of changes observed in the absence/presence of RcsB or upon induction of the RcsCDB system by mutations in IgaA. A representative example is that of the yciGFE genes, maximally expressed in the wild-type strain (i.e., not induced in the igaA1 mutant), but displaying an absolute dependence on both RcsB and RcsA for expression.

Another objective of this study was to analyze how the Rcs regulon could be affected by a decrease in the relative levels of RcsC and RcsD. To our knowledge, no precedent exists for this type of analysis at the genome level with reduced amounts of the sensor protein(s) of a two-component system. Colanic acid genes were induced, albeit to different extents, in three distinct mucoid serovar Typhimurium strains having augmented Rcs responses: MD0842-J4, MD0855-J10, and the igaA1 mutant (42). Based on this parameter, the intensity of the RcsCDB response was graded in an igaA1 mutant > MD0855-J10 > MD0842-J4 order. Interestingly, phenotypic traits performed with IgaA-deficient bacteria showed virulence attenuation only for the MD0855-J10 strain (Table 5). Since both the MD0842-J4 and MD0855-J10 strains upregulate colanic acid capsule genes and the yjbEFGH operon, we see as unlikely a negative effect of exopolysaccharide overproduction on virulence (28, 45). On the other hand, MD0842-J4 and MD0855-J10 displayed different levels of expression of the virulence genes spvAB, a change that could explain the observed phenotypes. Obviously, alterations in the expression and/or activities of other virulence factors may also contribute to the attenuation of the MD0855-J10 mutant. Overall, the genome expression data collected from the IgaA-deficient strains demonstrate that the strong Rcs responses resulting from the absence of this attenuator can be differentiated.

The regulation exerted on the spv genes by the RcsCDB system was also of interest. The transcriptional activity of spvA as registered in different genetic backgrounds revealed a positive regulation of RcsB on this virulence gene. It is noteworthy that spvA expression was found to be normal in rscC and rcsD mutants, which indicated that an intact RcsC→RcsD→RcsB phosphorelay was not required for this regulation. This observation, however, contrasted with the reduced spvA expression linked to the stimulation of the RcsCDB system. This finding prompted us to consider an involvement of dephosphorylated RcsB. The data obtained with the nonphosphorylatable RcsB(D56Q) variant demonstrated that RcsB does not need to be phosphorylated to promote spvA expression (Fig. 5B). To our knowledge, these results are the first evidence sustaining the role of dephosphorylated RcsB as a regulator of gene expression. Precedents for response regulators that retain activity independently of phosphorylation are nonetheless known for NblR in Synechococcus sp. (48), HP1043 in Helicobacter pylori (50), and FrsZ in Myxococcus xanthus (25). In our model, there is, however, a feature not described in the previous cases. RcsB(D56Q) causes a more-pronounced positive effect on spvA when it is produced in an rcsB⁺ background, which suggests that both RcsB∼P and dephosphorylated RcsB may be required for attaining proper spvA expression. This double requirement was tentatively integrated into a model in which RcsB∼P is proposed to regulate positively a hypothetical coregulatory protein that could act together with dephosphorylated RcsB on spv promoters (Fig. 6). This hypothetical RcsB-coregulator complex may act on the promoter of spvR, the spvABC operon, or both (58). Future work is required to discern which of these scenarios is correct. An additional unexpected finding was the low expression of spvA registered in glucose-containing medium compared to that under conditions in which glycerol was the only carbon source. Although the basis of these differences is currently unknown, this new level of regulation supports the idea of an intricate regulatory circuit operating on the spvABC genes. Besides the control exerted on these genes by RpoS (59), the dedicated regulator SpvR (58), polynucleotide phosphorylase (60), and the RcsCDB system (this study), catabolic repression may be relevant.

FIG. 6.

Model showing the putative role of nonphosphorylated RcsB in the positive regulation of spvA expression. This regulation could be hypothetically exerted on the spvABC regulatory region or, indirectly, via the dedicated activator SpvR. The model also highlights the requirement for this process of RcsB∼P, probably stimulating the production of an unknown regulatory factor (X) that may act as a coregulator together with dephosphorylated RcsB. The model also includes two alternate mechanisms of attenuation of the RcsC→RcsD→RcsB phosphorelay by IgaA, either by inhibition of autophosphorylation in the H1 kinase domain of RcsC or stimulation of the phosphatase activity associated with the D1 receiver domain of RcsC (dotted lines). This role of IgaA may be essential to fine-tune the switch between RcsB∼P and nonphosphorylated RcsB and, as a consequence, to dictate the intensity at which the Rcs regulon is expressed.

In summary, the diminished spvA expression observed in the igaA1 mutant can be a consequence of the high RcsB∼P/RcsB ratio existing upon stimulation of the RcsC→RcsD→RcsB phosphorelay. IgaA may then ensure proper spvA expression by facilitating the switch of RcsB∼P to the dephosphorylated state. In a more general role, IgaA could be responsible for constantly adjusting the RcsB∼P/RcsB ratio in response to the amount and/or type of signals perceived by the bacteria. However, how could IgaA ultimately perform this hypothetical function? Given its location as an integral inner membrane protein, IgaA has the potential to interact with membrane components of the Rcs system as the primary sensor RcsC. Following this, IgaA might attenuate the autophosphorylation taking place in the H1 kinase domain or, alternatively, stimulate the phosphatase activity at the receiver D1 domain of RscC (12, 26, 39) (Fig. 6). The possibility of IgaA also modulating the autophosphorylation rate of RcsB in the presence other phospho-donors, such as acetyl-phosphate (26), cannot be discarded with the available data. Work directed to explore these diverse mechanisms is in progress.