The Symbiosis Regulator CbrA Modulates a Complex Regulatory Network Affecting the Flagellar Apparatus and Cell Envelope Proteins

Abstract

Sinorhizobium meliloti participates in a nitrogen-fixing symbiosis with legume plant host species of the genera Medicago, Melilotus, and Trigonella. We recently identified an S. meliloti two-component sensory histidine kinase, CbrA, which is absolutely required to establish a successful symbiosis with Medicago sativa (K. E. Gibson, G. R. Campbell, J. Lloret, and G. C. Walker, J. Bacteriol. 188:4508-4521, 2006). In addition to having a symbiotic defect, the cbrA::Tn5 mutant also has free-living phenotypes that suggest a cell envelope perturbation. Because the bases for these phenotypes are not well understood, we undertook an identification of CbrA-regulated genes. We performed a microarray analysis and compared the transcriptome of the cbrA::Tn5 mutant to that of the wild type. Our global analysis of gene expression identified 162 genes that are differentially expressed in the cbrA::Tn5 mutant, including those encoding proteins involved in motility and chemotaxis, metabolism, and cell envelope function. With regard to those genes with a known role in symbiosis, we observed increased expression of nine genes with overlapping functions in bacterial invasion of its host, which suggests that the mutant could be competent for invasion. Since these CbrA-repressed genes are vital to the invasion process, it appears that down-regulation of CbrA activity is important at this stage of nodule development. In contrast, our previous work showed that CbrA is required for bacteria to establish themselves within the host as nitrogen-fixing symbionts. Therefore, we propose a model in which CbrA functions as a developmental switch during symbiosis.

As a model system for studying methods that invasive bacteria employ to transition from a free-living environment to their niche within the host, we explored how the gram-negative bacterium Sinorhizobium meliloti establishes a nitrogen-fixing symbiosis within the roots of leguminous plants such as Medicago sativa. These bacteria elicit formation of a specialized plant organ, called a nodule, within which they establish themselves as chronic intracellular symbionts. Once intimately associated with their host, these bacteria reduce dinitrogen to biologically accessible ammonia, which is provided to the host in exchange for carbon derived from photosynthesis. Thus, these symbiotic bacteria play a crucial role in the global nitrogen cycle and can reduce the need for synthetic nitrogen fertilizers (83).

Nodule formation is a highly regulated and complex developmental process (11). Nodulation is initiated by a highly specific chemical dialog involving the bacterial perception of plant flavonoids and the responding synthesis of Nod factor, which elicits nodule organogenesis (26, 30, 36, 45). Bacteria subsequently invade the newly formed nodule through a plant-derived structure called the infection thread (27). There are a number of complex polysaccharides secreted by S. meliloti that are required to promote infection thread formation and growth (26, 36). For instance, low-molecular-weight (LMW) forms of the exopolysaccharide succinoglycan (EPS I) promote infection thread invasion, and they provide an overlapping function with the exopolysaccharide EPS II and K antigen (18, 58, 59). Cyclic β-(1,2)-glucan production is also required for infection thread invasion, although this requirement can be bypassed through the overproduction of succinoglycan (23, 52).

Once bacteria gain entry into host tissues via the infection thread, they are individually taken up into a membrane-bound compartment within the host cell cytoplasm where they differentiate into physiologically and morphologically distinct bacteroid forms capable of nitrogen fixation (56, 62). As part of this developmental process, both the bacterium and its plant host undergo profound changes in gene expression (3). In comparison to the requirements for infection thread invasion, the bacterial requirements for intracellular adaptation and differentiation are less well understood. These events are complicated by physiological challenges imposed by the plant host, which include the release of reactive oxygen species and an acidic intracellular host compartment (6, 9, 63, 70). It is now well established that certain molecular mechanisms these chronic endosymbionts use to establish a beneficial relationship with the plant host are shared with the chronic pathogen species Agrobacterium and Brucella (5, 82). Therefore, any insights gained into the Sinorhizobium-Medicago symbiosis have potential to further our understanding of these phylogenetically related pathogens.

We recently identified a new regulatory protein, CbrA, whose activity is crucial to S. meliloti for establishing an effective symbiosis with the host legume M. sativa (32). CbrA is a putative two-component histidine kinase with a sensory PAS domain, a ligand-binding motif commonly involved in sensing redox status (80). In contrast to a successful symbiosis that produces pink nodules, the cbrA::Tn5 mutant primarily elicits the formation of immature nodules from which fewer bacteria can be recovered, suggesting that the mutant either is unable to colonize its host efficiently or is compromised for survival within the host environment. Moreover, the cbrA::Tn5 mutant is unable to compete successfully with the wild type for nodule colonization during a mixed infection.

We originally identified the cbrA::Tn5 mutant in a screen based on its succinoglycan overproduction phenotype (32). Succinoglycan is an exopolysaccharide that plays a critical role in infection thread development and hence in nodule invasion (18). Importantly, our identification of the genes exoH, exoK, and exoT as CbrA-regulated genes suggests that the cbrA::Tn5 mutant is biased toward production of succinylated LMW forms of this exopolysaccharide, the structural forms most effective at promoting bacterial invasion via the infection thread (58). It therefore seems unlikely that succinoglycan overproduction is responsible for the severe symbiotic defects of the cbrA::Tn5 mutant. However, the cbrA::Tn5 mutant also exhibits a severe sensitivity to a variety of membrane-disrupting agents, including the bile salt deoxycholate and the hydrophobic dye crystal violet (32), indicating that the mutant has a cell envelope defect that may account for the symbiotic phenotype of the mutant.

The physiology of the gram-negative cell envelope determines how these bacteria interact physically with their environment. In addition to its key role in regulating the flow of molecules into and out of the cell (54), the envelope modulates host cell adhesion and host responses to bacterial invasion through a variety of mechanisms (50, 53, 60). The well-characterized S. meliloti lpsB and bacA mutants have cell envelope defects which result from distinct alterations to the lipopolysaccharide (LPS) that are associated with bacterial sensitivity to a variety of toxic agents (13, 24). These mutants fail to establish an effective symbiosis due to aborted infections that allow bacterial uptake into the host cell cytoplasm but are blocked for bacteroid differentiation (13, 34). In fact, these mutants appear to degrade rapidly upon uptake into the host membrane-bound compartment, suggesting they are either unable to survive within this intracellular environment or they are subject to a lethal host innate immune response. In contrast with the LPS, little is known regarding the requirement for S. meliloti outer membrane proteins (OMPs) in either cell envelope integrity or symbiosis.

The cbrA::Tn5 mutation affects gene expression primarily during the stationary phase growth, indicating that CbrA activity is responsive to the growth phase (32). We therefore examined global gene expression in the cbrA::Tn5 mutant to address the role of CbrA in both stationary phase regulation and cell envelope physiology and anticipated that we would gain potentially novel insights into bacterial requirements for host colonization. The set of differentially expressed genes identified in this study is significantly enriched for those encoding OMPs, secreted extracellular proteins, and proteins involved in motility and chemotaxis. Our results suggest CbrA has a broad regulatory role in modulating the physiology of the bacterial outer membrane and cell surface and thereby regulates the expression of proteins that could be involved in direct contact with the host during symbiosis.

MATERIALS AND METHODS

Bacterial strains, phage, plasmids, and growth conditions.

Bacterial strains and a generalized transducing phage are listed in Table 1. Sinorhizobium meliloti strains were grown in LB/MC (LB medium supplemented with 2.5 mM MgSO₄ and 2.5 mM CaCl₂). Unless otherwise noted, S. meliloti strains were cultured at 30°C in LB/MC into stationary phase (approximate optical density at 600 nm [OD₆₀₀] of 5.0) for physiological and gene expression assays. Escherichia coli was grown in LB medium supplemented with appropriate antibiotics at 37°C. Antibiotic concentrations used for S. meliloti selection are as follows: streptomycin, 500 μg/ml; neomycin, 200 μg/ml; gentamicin, 12.5 μg/ml; and spectinomycin, 100 μg/ml. Antibiotic concentrations used for E. coli selection were as follows: spectinomycin, 50 μg/ml; chloramphenicol, 34 μg/ml; and kanamycin, 50 μg/ml. Symbiosis assays were performed as previously described (31), except that buffered nodulation medium was buffered at pH 6.5, and 1 mM aminoethoxyvinyl glycine (Sigma) was added for Medicago truncatula assays.

TABLE 1.

Bacterial strains, phage, plasmids, and PCR primers used in this study

Strain, phage, plasmid, primer	Relevant characteristics	Source/reference
Strain
DH5α	E. coli [endA1 hsdR17 supE44 thi-1 recA1 gyrA relA1 Δ(lacZYA-argG)U169 deoR]	BRL Corp.
MT616	E. coli MM294 pRK600, Cmr	T. Finan
Rm1021	SU47 Smr	F. Ausubel
KEG1002	Rm1021 ropB1′-uidA+	This study
KEG1003	KEG1002 cbrA::Tn5-233	This study
KEG2016	Rm1021 cbrA::Tn5	32
KEG2141	MB669 cbrA::Tn5	This study
KEG2143	MB670 cbrA::Tn5	This study
KEG2145	MB671 cbrA::Tn5	This study
KEG2147	MB672 cbrA::Tn5	This study
KEG2149	MB673 cbrA::Tn5	This study
KEG2150	Rm1021 bioS-uidA+	This study
KEG2151	KEG2150 cbrA::Tn5	This study
KEG2152	Rm1021 SMc00489-uidA+	This study
KEG2153	KEG2152 cbrA::Tn5	This study
KEG2155	VO2674 cbrA::Tn5	This study
MB669	Rm1021 flaC′-uidA+	This study
MB670	Rm1021 mcpX′-uidA+	This study
MB671	Rm1021 mcpU′-uidA+	This study
MB672	Rm1021 SMc00888′-uidA+	This study
MB673	Rm1021 SMc00887′-uidA+	This study
VO2674	Rm1021 nex18′-bacA+-uidA+	55
Phage
φM12	Generalized transducing phage	T. Finan
Plasmid
pCRBluntII-TOPO	Blunt PCR cloning vector	Invitrogen
pK18mobsacB	Allele exchange vector	71
pWM3	Vector with nonpolar uidA+ cassette	49
pVO155	Suicide vector with uidA+ cassette, Nmr	V. Oke
pVO345	Suicide vector with uidA+ cassette, Spr	V. Oke
pKEG11	pVO155 ropB1′-uidA+	This study
pKEG15	pK18mobsacB bioS′-uidA+	This study
pKEG16	pK18mobsacB SMc00489′-uidA+	This study
pMB694	pVO345 flaC′-uidA+	This study
pMB695	pVO345 mcpX′-uidA+	This study
pMB696	pVO345 mcpU′-uidA+	This study
pMB697	pVO345 SMc00888′-uidA+	This study
pMB698	pVO345 SMc00887′-uidA+	This study
PCR primer
bioS-Pst′1	CTGCAGGTCCCACAGGTAGCGTC	This study
bioS-BamHI′1	CGTAGGCGGATCCGGCAC	This study
bioS-BamHI′2	GTGCCGGATCCGCCTACG	This study
bioS-Pst′2	CTGCAGGGTCATCGGTCTCGGCTC	This study
flaC-Age	CTTGCGTCAGACTTGCCGGAT	This study
flaC-Xba	GCGCGCAATGTCTGCAACGCC	This study
mcpU-Age	AGCTCCCGCTAGCGTCCAGCC	This study
mcpU-Xba	ACGCGGCAACGAGAATGCGCG	This study
mcpX-Age	CTAGGGCACTTCCAGGAAAAG	This study
mcpX-Avr	GCCAGCGATCTGGCTTGGAAC	This study
ropB1-up	GCTCTAGAGCCGCAAACTCAATG	This study
ropB1-dn	CCTTCATCGTTTCGAGCAGGTTCC	This study
C00489-Pst′1	CTGCAGGAGGGCGTTC	This study
C00489-BamHI′1	CGTCGCGGATCCAGGTCGTAG	This study
C00489-BamHI′2	CTACGACCTGGATCCGCGACG	This study
C00489-Pst′2	CTGCAGGAACGTGTAGCCGAGCAT	This study
C00887-Age	GAACTCGGCAATGCACGCCAC	This study
C00887-Xba	CCAGCGGCAGGACGATGGCGA	This study
C00888-Age	CACTTCTCCATAGCTGTCGGC	This study
C00888-Xba	GATGTCTGCGCCGCCCGACGC	This study

Genetic techniques and DNA manipulations.

Bacteriophage transductions with φM12 (25) and triparental matings (44) were performed as previously described. S. meliloti genomic DNA was isolated as described previously (2). Plasmids and primer sequences are listed in Table 1. To construct the pK18mobsacB derivatives, genomic DNA was used as a template for PCRs with Thermalase DNA polymerase (Invitrogen). The PCR products were initially cloned and sequenced in pCRBluntII-TOPO (Invitrogen) and then subcloned into pK18mobsacB (71). The suicide vector pK18mobsacB was used for allelic replacement so that a gene of interest could be disrupted with the uidA⁺ transcription fusion cassette. The uidA⁺ cassette was obtained by BamHI digestion of pWM3 (49). The suicide vectors pVO155 and pVO345 were used to create uidA⁺ transcription fusions at native gene loci. Derivatives of the suicide vectors pVO155, pVO345, and pK18mobsacB were mated into strain Rm1021 by selecting for Sm^r Nm^r or Sm^r Sp^r, and the correct genomic location of the plasmid was verified through PCR. Rm1021 strains containing pK18mobsacB derivatives were cultured overnight in LB/MC medium, and an aliquot of the culture was plated onto LB medium containing 10% sucrose to promote the loss of the pK18mobsacB vector. The resulting colonies were screened for lack of growth on LB medium containing Nm. Nm^s Suc^r colonies were further screened by PCR to determine whether the native allele was interrupted with the uidA⁺ cassette. Those Nm^s Suc^r colonies with the uidA⁺ cassette properly integrated into the genome were retained for further study.

Affymetrix GeneChip analysis.

Design of the Symbiosis GeneChip, RNA isolation, cDNA synthesis, labeling, and hybridization were as described previously (4). Three biological replicates, 20-ml cultures grown in separate 250-ml flasks from single colonies, were used for each experimental strain (KEG2016) or control (Rm1021) strain data set. Experimental and control cultures were grown in LB/MC medium with streptomycin selection to similar OD₆₀₀ values ranging from 4.1 to 4.5. For data mining, we used Affymetrix software (GCOS and DMT) as described previously (4). Briefly, we normalized using all probe sets with the target signal value set to 500 and used comparison, also known as relative, expression analysis: one array is designated as experimental and the other as baseline. Hence, this analysis with three control and three experimental arrays would yield nine pairwise comparisons. We considered an average signal log ratio (SLR) of ≥1.0 for the increase or decrease change calls to be significant only if eight or nine of the nine pairwise comparisons were deemed by the software to be significantly changed (a P value of ≤0.05). Note that SLR is expressed as the log₂ ratio of the change, i.e., an SLR of 1 equals a twofold change. For data shown in Table 2, SLR values were converted to change (n = fold) values.

TABLE 2.

Functional grouping for differentially expressed genes identified by microarray

Gene	Gene product (reference)	SLR of cbrA::Tn5/WT converted to fold change values
Regulators
sinI (SMc00168)	Quorum-sensing autoinducer synthase (47)	+2.59
cspA5 (SMc00289)	Putative CspD family cold shock protein (43)	−2.44
SMc00887a,b	Putative c-di-GMP GGDFF/EAL enzyme	−2.58
SMc00888a,b	Putative hybrid histidine kinase with an N-terminal PAS domain and a C-terminal response regulator domain	−2.70
bioS (SMc02061)a,b	LysR family biotin-responsive transcription factor (37)	−3.74
SMc02369	Putative PleC-like histidine kinase with a PAS/PAC domain	−2.34
visN (SMc03015)	LuxR family transcriptional regulator of fla genes (75)	−2.44
rem (SMc03046)	OmpR family response regulator of fla genes (68)	−3.24
Exopolysaccharide biosynthesis
Succinoglycan biosynthesis and modification
exoY (SMb20946)	Galactosyltransferase (51)	+2.42
exoH (SMb20954)	Succinyltransferase (35)	+3.44
exoK (SMb20955)	Endo-β-1,3-1,4-glycanase (7)	+2.65
exoL (SMb20956)	Glucosyltransferase (64)	+2.20
exoA (SMb20957)	Glucosyltransferase (64)	+2.04
exoN (SMb20960)	Glucosyltransferase (8)	+2.45
exoW (SMb21690)	Glucosyltransferase (64)	+2.78
Additional polysaccharide biosynthesis and export
pssF (SMb20748)	Putative Wca-like type 2 glycosyltransferase	+2.82
SMc01793	Putative RfaG-like type 1 glycosyltransferase	−2.28
SMc01794	Putative Wza-like periplasmic polysaccharide transporter	−2.04
ndvA (SMc03900)	Inner membrane ABC-family transporter of cyclic β-(1,2)-glucan (78)	+3.28
Flagellar biosynthesis and motility (representative subset of 56 genes)
mcpU (SMc00975)a	Chemoreceptor (48)	−4.12
mcpX (SMc01104)a	Chemoreceptor (48)	−4.73
cheY1 (SMc03006)	Chemotaxis regulator protein (76)	−5.00
fliI (SMc03025)	Flagellum-specific ATP synthase (76)	−10.40
flaC (SMc03040)a	Flagellin protein C (76)	−3.98
motC (SMc03043)	Chemotaxis motility protein C (76)	−10.54
Histidine uptake and metabolism
SMa0387	Putative HisC-like histidinol-phosphate aminotransferase	+2.78
hutU (SMb21163)	Putative urocanate hydratase (urocanase)	+2.63
hutH1 (SMb21165)	Putative histidine ammonia-lyase	+2.54
hutI (SMb21166)	Putative imidazolonepropionase	+3.06
hisV (SMc00670)	HisV/HutV histidine transport ABC transporter cytoplasmic ATP-binding cassette (10)	+2.80
hisW (SMc00671)	HisW/HutW histidine transport ABC transporter inner membrane permease (10)	+3.04
hisX (SMc00672)	HisX/HutX histidine transport ABC transporter periplasmic binding protein (10)	+3.49
SMc00673	Putative hydrolase	+3.06
Chaperones and proteolytic processing
SMb21295	Putative IpaA-like Hsp20 chaperone	+3.52
SMc00360	Putative periplasmic transglutaminase-like cysteine protease (DUF920)	−2.32
SMc00539	Putative M23 family membrane-bound metaloprotease (COG4942)	−2.18
SMc00998	Putative periplasmic transglutaminase-like cysteine protease (DUF920)	−2.72
SMc02825	Putative M17 family aminopeptidase	−2.44
SMc04040	Putative IpaA-like Hsp20 chaperone	+3.19
Pilus assembly
cpaF1 (SMc02820)	Putative cytoplasmic pilus assembly protein	−2.12
SMc04059	Putative TadG-like pilus assembly protein	−2.23
cpaE1 (SMc04109)	Putative pilus assembly protein	−2.29
cpaC1 (SMc04111)	Putative outer membrane pilus assembly protein	−2.04
Outer membrane proteins
Opacity-related subgroup of OmpA-like proteins
SMc00141	Putative OMP associated with a DUF612 domain	+5.70
SMc00489a,b	Putative Pseudomonas aeruginosa PagL lipid A 3-O-deacylase ortholog (69)	−2.31
ropB1 (SMc00604)a	Putative Rhizobium leguminosarum RopB ortholog (14)	+2.36
SMc00638	Putative Mesorhizobium huakuii Opa22 ortholog	−2.39
SMc00639	Putative Mesorhizobium huakuii Opa22 ortholog	−2.05
Miscellaneous OMP tspO (SMa1079)	Rhodobacter sphaeroides TspO homolog required for nutrient deprivation regulation of ndiAB (21)	+3.13
SMc00986	Putative OMP with two copies of the DUF1217 domain	−2.90
Secreted extracellular proteins nex18 (SMa1077)a	Fasciclin-like protein (55)	+2.61
SMb20838	Putative NodO-like hemolysin Ca2+-binding metalloprotease	+3.74
SMc03108	Putative NodO-like hemolysin Ca2+-binding metalloprotease	+3.30
SMc04236	Putative glycine-rich outer membrane autotransporter protein	+3.95

^aDifferentially expressed gene whose regulation was confirmed with a gene fusion.

^bDifferentially expressed gene whose symbiotic phenotype was assayed; each mutant assayed was found to promote wild-type symbiosis with M. sativa and M. truncatula.

β-Glucuronidase activity assays.

The relative expression level of each uidA⁺ transcription fusion was determined by assaying β-glucuronidase activity as previously described (39) with the following modifications. The assay buffer was supplemented with 10 mM EDTA (pH 8.0) and 0.1% sarcosyl, and the enzyme assays were performed with 5 mM p-nitrophenyl β-d-glucuronide substrate (Sigma). At specific time points, the reaction was stopped and the sample was centrifuged for 5 min at 18,000 × g to remove cellular debris. p-Nitrophenol absorbance of the clarified supernatant was measured at 415 nm, and modified Miller units were determined with the following formula: 1,000(A₄₁₅/tvp), where t is the reaction time in minutes, v is the volume of culture assayed in ml, and p is the concentration of total cell protein present in the reaction. We isolated total cell protein by the addition of 0.2 volume of 100% trichloroacetic acid (TCA), followed by centrifugation for 10 min at 18,000 × g. The resulting TCA pellet was washed with 100% acetone, dried, and resuspended in 50 mM Tris (pH 8). Protein concentrations were determined by a Bradford assay (Bio-Rad). For each strain assayed, the average and standard deviation were derived from four independent cultures.

Isolation and analysis of flagella.

S. meliloti cultures were vortexed vigorously three times for 15 s in order to shear flagellar filaments from the cell surface. The vortexed culture was centrifuged for 30 min at 3,500 × g, and the cleared supernatant was isolated. Flagellar filaments within the supernatant were further isolated by ultracentrifugation at 40,000 × g for 40 min. Protein in the flagellar pellet was solubilized in 50 mM Tris (pH 7.5) buffer containing 250 mM NaCl and 10% glycerol, with a volume of buffer normalized for total protein concentration of the originating culture. We isolated total protein from an aliquot of each vortexed culture by the addition of 0.2 volume of 100% TCA, followed by centrifugation at 18,000 × g at 4°C for 10 min. The resulting TCA pellet was washed with 100% acetone, dried, and resuspended in 50 mM Tris (pH 8). Protein concentrations were determined by a Bradford assay (Bio-Rad). The flagellar preparation was analyzed by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Flagellar protein was detected with Coomassie stain, and the resulting band was isolated and analyzed by matrix-assisted laser desorption ionization-time of flight at the MIT Biopolymers Core Facility to ensure it represented flagellar protein. A 1:25 dilution of the flagellar preparation was again subjected to 12% SDS-PAGE, and protein was detected using quantitative Sypro orange dye according to the manufacturer's instructions (Molecular Probes). The gel was scanned using a Typhoon 9400 model imager (Amersham Biosciences), and protein bands were quantified relative to an internal standard of bovine serum albumin (Sigma).

Bioinformatics analyses of differentially expressed genes.

For those differentially expressed genes with no known function, we searched for homologous functional or conserved domains using HHPred (http://toolkit.tuebingen.mpg.de/hhpred), which uses both sequence identity and predictions of secondary structure in the query to search several protein databases. We used the PSORT algorithm (http://www.psort.org/) to predict protein localization. We identified the closest characterized PAS domain family member for those PAS domains present in CbrA, SMc02369, and SMc00888 by performing database searches with both HHPred and Phyre (http://www.sbg.bio.ic.ac.uk/phyre/), which is also a protein structure prediction program that aligns a query with related domains of a known tertiary structure and highlights functional residues within the alignment. These secondary structure-based searches were more successful at identifying PAS domain relatives than a protein BLAST search (http://www.ncbi.nlm.nih.gov/BLAST/) because PAS domain relatives share more structural identity than sequence identity (61). The sequence alignment of E. coli OmpA and S. meliloti OmpA family members was performed based on predictions of secondary structure using HHPred and Phyre and the known structure of the membrane-spanning domain of OmpA (57).

RESULTS

Microarray analysis identifies potential CbrA-regulated genes during free-living stationary phase growth.

CbrA is a putative two-component sensory histidine kinase that is required to establish an effective symbiosis between Sinorhizobium meliloti and the host legume Medicago sativa (32). Notably, we found that CbrA-dependent effects on gene expression are observed primarily during stationary phase growth rather than exponential growth. In order to achieve a comprehensive identification of CbrA-regulated genes, we performed a global analysis using Affymetrix GeneChips to compare the pattern of gene expression in the cbrA::Tn5 mutant to that of the wild type under stationary phase growth conditions. Overall, the microarray experiment identified 80 genes with significantly increased expression (see Table S1 in the supplemental material) and 82 genes with significantly decreased expression (see Table S2 in the supplemental material) in the cbrA::Tn5 mutant, compared to those in the wild type. Nearly half of these genes have no predicted function, and this is particularly true of the most strongly induced genes in the cbrA::Tn5 mutant (see Table S1 in the supplemental material). However, some of the genes identified did cluster into functional groups (Table 2).

We performed a preliminary validation of the microarray experiment by comparing the expression pattern of succinoglycan biosynthesis genes (Table 2) from this study to previous data (32). Consistent with our previous results, we observed a threefold increase in the expression of the succinoglycan modification genes exoH and exoK in the cbrA::Tn5 mutant. However, the overall pattern of exo gene expression observed with the microarray highlighted a slightly different subset of exo genes than we found in our previous studies using transcription fusions and reverse transcriptase PCR. This may be attributable to significant differences in culture conditions (batch growth in culture tubes versus flasks) or perhaps to the differential sensitivity of the methods used. For instance, previously, we found dramatic CbrA-dependent regulation of exoT; however, its expression was too low to be reliably detected by GeneChip analysis. We also found that exoB levels were unchanged, in contrast with our previous results, and that several additional exo genes were subject to CbrA regulation (including the genes exoA, exoL, exoN, exoW, and exoY).

Interestingly, our expression data suggest that CbrA may modulate the production or transport of other polysaccharides, in combination with succinoglycan, that promote infection thread invasion. The cbrA::Tn5 mutant exhibits a threefold increase in expression of ndvA (Table 2), which encodes an inner-membrane ABC-type transporter of the cyclic β-(1,2)-glucan critical for infection thread invasion (23, 52, 78). There may also be an increase in expression of sinI (Table 2), which encodes a quorum-sensing autoinducer synthase whose activity is required to establish symbiosis in the absence of succinoglycan via its role in regulating ExpR-dependent EPS II production (46).

CbrA affects the expression of multiple regulators during stationary phase.

Our microarray analysis indicated that CbrA affects the expression of eight regulatory proteins and may therefore be situated at the top of a complex regulatory network (Table 2). For instance, the cbrA::Tn5 mutant exhibits decreased expression of bioS (Table 2), which encodes a LysR-like transcription factor whose expression increases in response to biotin addition and during the transition from exponential growth to stationary phase growth (37, 79). While BioS autoregulation was found to contribute to biotin-dependent induction of bioS, the factor required for stationary phase regulation of bioS was not identified previously. We performed an allelic exchange and replaced bioS⁺ with a bioS′-uidA⁺ fusion that serves as a reporter for bioS promoter activity and disrupts bioS function, eliminating the effects of autoregulation. During stationary phase, we observed a fourfold decrease in bioS′-uidA⁺ expression in the cbrA::Tn5 mutant compared to that in the wild type (Fig. 1, Table 2). In contrast, during exponential growth, there was no significant difference in the bioS expression of the wild type and that of the cbrA::Tn5 mutant (Fig. 1). Thus, CbrA regulates bioS in a growth phase-dependent manner and is required for full stationary phase induction of bioS. As reported previously (79), we found that a bioS null mutant is symbiotically proficient on M. sativa and is able to compete effectively with a bioS⁺-carrying strain for nodule colonization; this observation is also true of the related diploid host Medicago truncatula (data not shown).

FIG. 1.

CbrA plays an important role in growth phase regulation of bioS. β-Glucuronidase activity expressed from a bioS′-uidA⁺ fusion was measured from either exponential phase or stationary phase wild-type and cbrA::Tn5 cultures. The uidA⁺ transcription fusion serves as a reporter for promoter activity and also disrupts gene function.

The cbrA::Tn5 mutant also exhibits decreased expression of a putative operon encoding two regulatory factors: one is an uncharacterized two-component histidine kinase with a PAS domain, SMc00888, and the other, SMc00887, is a putative enzyme involved in c-di-GMP [bis-(3′,5′)-cyclic dimeric GMP] metabolism. The secondary signaling molecule, c-di-GMP, has recently been found to regulate complex bacterial developmental processes in diverse bacterial species (12, 66). The intracellular concentration of c-di-GMP is modulated through the enzymatic activity of proteins containing a GGDEF domain, which functions as a diguanylate cyclase, or an EAL domain that functions as a phosphodiesterase. We were therefore intrigued to find that the expression of SMc00887, which encodes a protein containing an N-terminal GGDEF and a C-terminal EAL domain, is decreased fourfold in the cbrA::Tn5 mutant (Fig. 2, Table 2). However, SMc00887 contains an unorthodox GGDEF motif in which the canonical glycine residues are replaced with alanine and glutamate (ADDEF), suggesting that this domain is enzymatically inactive (42). The GGDEF domain of SMc00887 may instead function as a sensory motif, perhaps of GTP levels, that regulates the phosphodiesterase activity of the associated EAL domain. Thus, we predict that decreased expression of SMc00887 may lead to increased c-di-GMP levels in the cbrA::Tn5 mutant as a result of lower cellular phosphodiesterase activity, although this hypothesis remains to be tested directly.

FIG. 2.

CbrA regulates the expression of genes predicted to affect c-di-GMP signaling. β-Glucuronidase activity was measured from stationary phase wild-type and cbrA::Tn5 cultures. The uidA⁺ transcription fusion present in each strain is indicated below the graph. The SMc00888′-uidA⁺ fusion was integrated in order to leave the gene intact while the SMc00887′-uidA⁺ fusion disrupted the native gene.

The start codon of SMc00887 overlaps with the stop codon of the upstream gene SMc00888 (ATGA), suggesting that the two genes are translationally coupled and perhaps functionally related. Similar to SMc00887, the expression of the SMc00888 locus is decreased threefold in the cbrA::Tn5 mutant during stationary phase growth (Fig. 2 and Table 2). SMc00888 encodes an uncharacterized hybrid two-component histidine kinase associated with an N-terminal sensory PAS domain and a C-terminal response regulator domain. Based on a homology search of all previously characterized PAS domains, the SMc00888 PAS domain (amino acid residues 142 to 239) is most similar to that of the S. meliloti oxygen sensor FixL (18% identical) and the Escherichia coli redox-sensitive phosphodiesterase DOS (17% identical), both of which bind a heme ligand (33). Curiously, the CbrA PAS domain (residues 729 to 880) also shares similarity to the FixL (27% identity) and DOS (21% identity) PAS domains. Given the important role that c-di-GMP can play in various pathogenic bacterium-host interactions, we tested the possibility that SMc00887 and SMc00888 could also participate in the regulation of symbiotic development. However, a strain lacking both SMc00888 and SMc00887 was symbiotically competent on both hosts M. sativa and M. truncatula (M. J. Barnett and S. R. Long, data not shown). Therefore, if these regulators do play a role in symbiosis, it is not an essential one, at least under laboratory conditions.

The expression of an uncharacterized two-component histidine kinase with a sensory PAS/PAC domain, SMc02369, is also decreased in the cbrA::Tn5 mutant (Table 2). The SMc02369 PAS motif (residues 125 to 389) contains a repeat similar to that of the Drosophila melanogaster PERIOD circadian clock protein (residues 2 to 297), which is involved in mediating protein-protein interactions (88), with these PAS motifs sharing 12% identity. Interestingly, of those histidine kinases with a known function, SMc02369 most closely resembles that of PleC of Caulobacter crescentus (29% identical, 62% similar). In C. crescentus, pleC mutants have paralyzed flagella and defects in pilus assembly (74, 84). As discussed below, the cbrA::Tn5 mutant also exhibits decreased motility and lowered expression of pilus biosynthesis genes, which may be attributable to altered SMc02369 expression.

CbrA modulates flagellar biosynthesis and cellular motility.

Flagellar motility is a highly regulated bacterial behavior that is responsive to nutritional availability, population density, and a variety of environmental stimuli (72, 77). In S. meliloti, there is a pattern of decreased flagellar motility under growth conditions or in mutants that increases succinoglycan production (3), and this link was also observed with the cbrA::Tn5 mutant. The largest functional clustering of genes with differential expression in the mutant compared to that in the wild type is comprised of those encoding flagellar biosynthesis and chemotaxis functions (56 genes or 35% [Table 1 and see Table S2 in the supplemental material]). The expression of these genes is significantly decreased in the cbrA::Tn5 mutant, ranging from 2- to 10-fold.

The expression of two flagellar biosynthesis regulators, visN and rem (68, 75), appears to be decreased in the cbrA::Tn5 mutant (Table 2). In S. meliloti, flagellar genes are expressed in a temporal hierarchy: class I genes are LuxR-type regulators encoded by visN and visR (75). Expression of all class II (for instance the flagellar assembly gene flaM) and class III (for instance the flagellar filament genes flaA through flaD) genes is strictly visN and visR dependent (75). Our microarray results indicate that the expression of visN, but not visR (data not shown), is decreased twofold in the cbrA::Tn5 mutant. Expression of another regulatory gene in the flagellar cluster, rem, is decreased approximately threefold in the cbrA::Tn5 mutant. Rem is an orphaned OmpR-type response regulator whose expression is visNR dependent and whose function is also required for expression of class II and class III flagellar genes (68). Given that VisN and Rem positively regulate motility and chemotaxis, their down-regulation in the cbrA::Tn5 mutant is consistent with the phenotypic decrease in motility discussed below. Therefore, the motility defect of the cbrA::Tn5 mutant is likely derived from the cumulative effect of multiple regulatory factors and is probably indirect.

We tested our microarray results using uidA⁺ transcription fusions present at the native gene loci and confirmed that expression of the flagellin gene flaC and the chemotaxis genes mcpU and mcpX is significantly decreased in the cbrA::Tn5 mutant (Fig. 3A and Table 2). To assess the overall impact of CbrA-dependent regulation on flagellar biosynthesis, we isolated flagellar filaments from cultures grown to stationary phase. We analyzed purified flagellar filament protein by SDS-PAGE and quantified the relative amounts of flagellin isolated from wild-type and cbrA::Tn5 cultures. Although we found flagella associated with the cbrA::Tn5 mutant, the abundance was 7- to 10-fold lower than that from the wild type (Fig. 3B). We also assayed motility by measuring bacterial swarming efficiency on LB/MC medium containing 0.22% agar. The swarm diameter formed by the cbrA::Tn5 mutant (2.8 ± 0.1 cm) was significantly decreased relative to that formed by the wild type (4.6 ± 0.1 cm) after 3 days (Fig. 3C). Although flagellar motility is not strictly required for nodulation, mutants defective in flagellar chemotaxis are less competitive at colonizing sites of nodule initiation on the root surface (1). Thus, the cbrA::Tn5 mutant motility defect could account, at least in part, for the inability of the mutant to compete with the wild type for nodule colonization.

FIG. 3.

CbrA regulates flagellar motility and chemotaxis. (a) β-Glucuronidase activity was measured from stationary phase wild-type and cbrA::Tn5 cultures. The uidA⁺ transcription fusion present in each strain, indicated below the graph, was integrated into the genome in order to leave gene function intact. (b) Flagellar filaments isolated from wild-type and cbrA::Tn5 stationary phase cultures were analyzed by SDS-PAGE and Coomassie protein detection. (c) The swimming ability of wild-type and cbrA::Tn5 bacteria was assessed after 3 days' incubation on LB/MC medium supplemented with 0.22% agar.

Another cell surface structure whose biosynthesis may be regulated by CbrA is the type IV pilus, composed of PilA subunits. Although expression of pilA1 was unaltered, the expression of four genes required for pilus assembly was decreased in the cbrA::Tn5 mutant (Table 2). This cluster of pilus assembly genes closely resembles that of Caulobacter crescentus cpaA-cpaF, which is required for cell cycle-regulated biosynthesis of polar pili (73).

CbrA affects the expression of genes predicted to encode outer membrane and secreted extracellular proteins.

The cbrA::Tn5 mutant exhibits an envelope defect as reflected by its severe sensitivity to membrane-disrupting agents (32), so we looked closely at those differentially expressed genes predicted to encode extracytoplasmic envelope proteins. We were also interested in the regulation of genes encoding outer membrane and secreted extracellular proteins because these factors have the potential to be involved in direct communication with the host during symbiosis.

Based on the PSORT algorithm, we estimate that 28% of the genes differentially expressed between the cbrA::Tn5 mutant and the wild type encode cytoplasmic proteins (Fig. 4b), similar to the proportion of the genome predicted to encode cytoplasmic proteins (Fig. 4a). However, 27% of these differentially expressed cytoplasmic factors are involved in biosynthetic processes that ultimately affect cell surface physiology, such as the exo genes, and another 47% are involved in flagellar motility. We estimate that 6% of differentially expressed genes encode outer membrane proteins and another 6% encode secreted extracellular proteins. In contrast, only 1.2% of the genome is predicted to encode secreted outer membrane and extracellular proteins. Therefore, genes predicted to encode these secreted proteins are 10-fold enriched within our set of differentially expressed genes, indicating that an important function of CbrA is the regulation of bacterial outer-membrane and cell surface physiology.

FIG. 4.

Genes encoding secreted outer-membrane and extracellular proteins are highly enriched within the pool of CbrA-regulated genes. The subcellular localization of each protein was predicted using the PSORT algorithm. (a) The percentage of S. meliloti proteins predicted to be localized to each subcellular compartment or secreted extracellularly. (b) The percentage of genes differentially expressed in the cbrA::Tn5 mutant whose protein products are known or predicted to be localized to each subcellular compartment or secreted extracellularly. A few genes discussed in the text have been highlighted.

Given that little is known regarding the role of OMPs during either stationary phase adaptation or symbiosis, we were interested to find that the expression of several genes encoding known or putative OMPs was altered in the cbrA::Tn5 mutant, including ropB1 and tspO (Table 2). Like the product of ropB1, most of these OMPs belong to the opacity-related subgroup of OmpA-like proteins. Similar to OmpA, these OMPs are predicted to fold into an eight-stranded antiparallel β-barrel. However, sequence alignment based on secondary structure predictions suggests that none of these S. meliloti proteins contains the soluble OmpA periplasmic peptidoglycan-binding domain located at the C terminus, indicating that these OMPs likely impart an alternative function to the cell envelope. In particular, alignment of SMc00638 and SMc00639 with OmpA predicts significantly larger cell-surface-exposed loops in these S. meliloti OMPs (data not shown), suggesting the possibility of an extracellular function.

Expression of ropB1 is decreased dramatically in bacteroids compared that in free-living bacteria (4). Results from our microarray and uidA⁺ transcription fusion analyses indicate that ropB1 is also influenced by CbrA, such that the cbrA::Tn5 mutant exhibits a twofold increase in ropB1 expression compared to that in the wild type (Fig. 5 and Table 2). Although ropB1 is not required for symbiosis in S. meliloti (14), it is possible that unregulated or increased expression of this OMP could have a detrimental impact on symbiotic development. SMc00489 encodes another differentially expressed OmpA family member, and a SMc00489′-uidA⁺ fusion confirmed that expression is decreased twofold in the cbrA::Tn5 mutant relative to that in the wild type (Fig. 5 and Table 2). The protein encoded by SMc00489 has homology to PagL (69), which functions as a lipid A 3-O-deacylase and modifies the acylation pattern of LPS within the outer membrane (81). Since PagL function contributes to the virulence of Salmonella enterica serovar Typhimurium (40), we tested whether SMc00489 contributes to nodule development and symbiosis. The SMc00489′-uidA⁺ fusion disrupts gene function, so we tested the nodulation ability of this mutant with legume hosts M. sativa, M. truncatula, and Melilotus officinalis. We observed no obvious symbiotic defect: the SMc00489′-uidA⁺ mutant elicited the development of pink nitrogen-fixing nodules that support wild-type levels of plant growth, and the mutant was able to compete effectively with SMc00489⁺ bacteria for nodule occupancy (data not shown).

FIG. 5.

CbrA regulates the expression of genes encoding outer membrane proteins. β-Glucuronidase activity was measured from stationary phase wild-type and cbrA::Tn5 cultures. The uidA⁺ transcription fusion present in each strain is indicated below the graph. The SMc00489 fusion disrupts gene function while the ropB1 fusion leaves the gene intact.

We identified several genes encoding known or putative secreted extracellular proteins as differentially expressed in the cbrA::Tn5 mutant (Table 2), including the well-characterized succinoglycan modification gene exoK (89) and nex18 (55). The ExoK endoglycanase promotes bacterial invasion of host tissues during infection thread growth by converting high-molecular-weight forms of succinoglycan to the symbiotically active low-molecular-weight forms (58, 89). The nex18 locus encodes a secreted extracellular protein with sequence similarity (37%) to the Mycobacterium tuberculosis protein MPB70, which has a structural similarity to the C-terminal FAS1 domain of the cell adhesion protein fasciclin I and was postulated to mediate physical interaction between invading bacterial cells and the extracellular matrix of the host (16). In S. meliloti, nex18 is likely expressed during nodule invasion because its promoter is sufficient to drive in planta bacA expression for normal symbiosis (55). A nex18 null mutant was subsequently found to play a role in promoting nodule development: nearly half of the plants assayed were stunted for growth and displayed reduced nitrogen fixation, compared to those of the wild-type symbiosis (55). We utilized the nex18′-bacA⁺-′uidA⁺ fusion to further examine CbrA-dependent regulation of nex18 and found, despite the low basal β-glucuronidase activity derived from this fusion, that the cbrA::Tn5 mutant consistently expressed twofold more β-glucuronidase activity (2.0 ± 0.4 Miller units) than the wild type (0.9 ± 0.15 Miller units). Thus, CbrA affects the expression of extracellular proteins that contribute to successful host invasion and nodule development. The identification of differentially expressed genes in the cbrA::Tn5 mutant brought to light a few uncharacterized extracellular proteins that may also influence bacterial invasion. We identified SMc03108 and SMb20838 as differentially expressed genes with increased expression in the cbrA::Tn5 mutant (Table 2). These two genes are predicted to encode secreted extracellular proteins of the RTX (repeat in toxin) family that contribute to the pathogenesis of a wide variety of virulent bacteria (20, 41). The Rhizobium leguminosarum nodulation protein NodO, another member of the RTX family, was previously shown to play a positive role in promoting invasion of R. leguminosarum during infection thread growth (31, 85). It will therefore be of interest to test whether SMc03108 and SMb20838 play a similar role during S. meliloti invasion.

DISCUSSION

CbrA is a regulatory factor whose function is critically important for establishing an effective symbiosis between S. meliloti and M. sativa (32). The cbrA gene encodes a putative two-component histidine kinase containing a sensory PAS domain. In our original study, we observed that a Tn5 disruption within cbrA leads to altered stationary phase expression of several exo genes involved in succinoglycan production (32). This result suggested that CbrA could play an important role in regulating stationary phase physiology in S. meliloti, which lacks any gene encoding the RpoS sigma factor critical for the regulation of stationary phase physiology in enteric bacteria (28, 38). In order to gain a more comprehensive understanding of CbrA-dependent regulation, we used microarray analysis to identify additional genes differentially expressed between the cbrA::Tn5 mutant and the wild type during stationary phase growth.

The results presented here further indicate that CbrA affects gene expression primarily during stationary phase growth conditions. For instance, we show that stationary phase induction of bioS expression is primarily dependent on CbrA. Although little is known about the targets of BioS regulation, it was observed that bioS function contributes significantly to the resumption of growth during the exit from the stationary phase (79). CbrA-mediated induction of bioS expression may therefore promote adaptive physiology in cells growing under stationary phase conditions. Global gene expression studies have suggested that S. meliloti growth is relatively rapid while the bacteria are contained within the infection thread but that cellular growth slows considerably once bacteria are located within the host cell cytoplasm (15). Consistent with this hypothesis is the observation that the stringent response regulator RelA is required for symbiosis (86). It has also been suggested that stationary phase physiology contributes to successful adaptation to the intracellular host environment in the closely related Brucella spp. (29, 65, 67), which also contain a CbrA ortholog and lack a candidate rpoS gene. One outstanding issue for future study is the precise nature of the growth phase signal that modulates CbrA activity, as this stimulus may also contribute to the developmental progression of the S. meliloti symbiont during nodule invasion and differentiation.

In addition to bioS, there are seven regulatory factors whose expression is responsive to CbrA activity. These include the SMc00887 enzyme, predicted to affect levels of the secondary signaling molecule c-di-GMP, the two-component histidine kinase genes SMc00888 and SMc02369, and cspA5 and sinI (Table 2). Disruptions in several of the regulatory genes identified in this microarray screen, including bioS (79), sinI (46, 47), SMc00887, and SMc00888, have been tested for symbiosis and found to result in no developmental defect. This is presumably also true of the flagellar regulators visN and rem, since flagella are not strictly required for symbiosis (1). The cbrA::Tn5 mutant phenotype may therefore result from the misexpression of genes that are direct targets of the CbrA two-component pathway. Alternatively, the cbrA::Tn5 symbiosis defect may derive from the combined effect of this mutation on several downstream regulatory pathways that control redundant, or overlapping, functions. Detailed studies with each of these regulators will be required to uncover the organizational structure of this network and to determine whether individual genes differentially expressed in the cbrA::Tn5 mutant are a direct or indirect target of the CbrA two-component pathway. This regulatory circuit is further complicated in that it contains three orphan histidine kinases, CbrA, SMc00888, and SMc02369. Curiously, each of these histidine kinases contains a sensory PAS domain that may monitor cellular redox status, suggesting that redox changes could play an important role in determining the physiological output of this particular regulatory circuit.

During early infection, bacteria are exposed to plant-derived reactive oxygen species but later find themselves in a microaerobic environment that is required for nitrogen fixation. Thus, the environment of the infection thread is quite different from that of the host cell cytoplasm, and we were intrigued by the possibility that CbrA may function as a developmental redox sensor able to distinguish between these two niches. Taken together, our results are consistent with the model in which cbrA::Tn5 bacteria are locked into a physiological state that is specialized for early infection thread invasion but are subsequently unable to properly differentiate within the host cell cytoplasm. For instance, the mutant has increased expression of the genes exoH, exoK, and ndvA, which play key roles in modifying or transporting the succinoglycan and cyclic β-(1,2)-glucan required for infection thread invasion. There may also be an increase in expression of sinI, which regulates ExpR-dependent EPS II production (46). It will therefore be interesting to test whether the genes encoding Nex18 and the NodO-like RTX factors SMc03108 and SMb20838, which also exhibit increased expressed in the cbrA::Tn5 mutant, specifically contribute to infection thread growth.

Comparatively little is known about the requirements for bacterial adaptation and differentiation after cells are taken up into the host cell cytoplasm. However, cbrA::Tn5 mutant bacteria appear to be unable to induce proper nitrogenase gene expression within the nodule, based on the level of β-glucuronidase activity expressed from a nifH′-uidA⁺ fusion (K. E. Gibson and G. C. Walker, unpublished data), indicating that bacteroid differentiation is somehow perturbed. We find that CbrA affects the expression of several genes predicted to encode secreted outer membrane proteins. In fact, these genes are enriched within the set of differentially expressed genes identified in our microarray analysis. We verified CbrA-dependent regulation of both ropB1 and SMc00489, whose products closely resemble E. coli OmpA, with the exception that they lack the periplasmic peptidoglycan-binding domain. Although neither ropB1 nor SMc00489 is required for symbiosis, it is possible their proteins play a role in free-living stationary phase physiology. It will be of interest to determine whether other CbrA-regulated OmpA homologs play a role in either stationary phase physiology or nodule development. For instance, the OMPs encoded by SMc00638 and SMc00639 share a high degree of similarity (65% and 61%, respectively) to the Opa22 protein of Mesorhizobium huakuii, which is absolutely required for nodule formation (17).

Interestingly, the cbrA::Tn5 mutant shares several phenotypic characteristics with the exoS96::Tn5 mutant, which is altered in the expression and activity of the ExoS two-component histidine kinase (19): both are succinoglycan overproducers with motility defects (32, 87). In fact, a variety of growth conditions lead to increased succinoglycan production and a concomitant decrease in flagellar motility, suggesting that the regulation of these phenotypes is closely linked (3). Perhaps the tight regulatory link between succinoglycan production and flagellar motility is required to coordinate a complex developmental program such as biofilm formation in the soil or rhizobial infection thread invasion of its host. Regardless, the cbrA::Tn5 mutant has a severe sensitivity to membrane-disrupting agents that is not observed with the exoS96::Tn5 mutant (K. E. Gibson and G. C. Walker, unpublished observations). Similarly, the cbrA::Tn5 mutant has a dramatic nodule development defect while the exoS::Tn5 mutant elicits formation of pink, nitrogen-fixing nodules (22, 32). Therefore, the apparent envelope defect of the cbrA::Tn5 mutant may contribute significantly to the nodule development defect of the mutant. We will continue to dissect the roles CbrA-regulated OMPs play in both free-living stationary phase physiology and symbiotic development, individually and in combination, as a means of gaining a better understanding of the function of the cell envelope during various stages of symbiotic development.