INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Competition among microorganisms determines the outcome of many biological events in nature, and yet competitiveness is poorly understood. Lack of knowledge of the mechanistic basis for competitiveness is due, in large part, to the difficulty in conducting a genetic analysis of this quantitative trait. Identification of mutants affected in competitiveness is challenging in many microbial systems due to variability and the need for a high degree of replication in mutant screens (26, 32).

The Rhizobium-legume symbiosis provides a good model system with which to study the molecular basis of bacterial competitiveness, because nodulation competitiveness is a readily quantifiable trait. Rhizobial species establish mutualistic relationships with specific leguminous plants by initiating the development of a specialized plant organ, the root nodule, in which the bacteria fix atmospheric nitrogen. The ability of a particular rhizobial strain to establish this in the presence of other strains is known as nodulation competitiveness. Nodulation competitiveness is measured by comparing the proportions of rhizobial strains that are initially applied to the seed with the proportion of the nodules that are later occupied by each strain.

The Rhizobium etli-bean symbiosis is well suited to the study of the genetic basis of nodulation competitiveness. Variability can be minimized by using a genetically homogeneous plant host population, such as common bean (Phaseolus vulgaris), which is highly inbred. Moreover, the nodules usually contain a pure culture of the successful competitor, which distinguishes this microbial competition from many others in which detection of competitive success is difficult.

The rosR gene likely encodes a regulator that plays a critical role in both nodulation competitiveness and determination of cell surface characteristics in R. etli (2, 10). A rosR mutant was originally identified by its distinctive domed colony morphology that results from its hydrophobic cell surface. The rosR mutant nodulates and fixes nitrogen, but when the mutant and the parent are coinoculated in equal amounts, nearly all of the root nodules are occupied by the parent strain. A vast (approximately 17,000-fold) excess of the mutant is required to achieve equal nodule occupancy, indicating that the rosR mutant is drastically reduced in nodulation competitiveness.

RosR is 80% identical to MucR from Sinorhizobium meliloti and Ros from Agrobacterium tumefaciens (16, 24). Ros and MucR act as transcriptional repressors by binding DNA sequences in the promoter regions of regulated genes via putative zinc finger motifs (15, 24). MucR affects the production of an alternative exopolysaccharide (EPS), EPS II (galactoglucan), in place of the normal EPS I (succinoglycan) by repressing transcription of the genes involved in EPS II synthesis (24, 39). Ros represses the virC and virD operons in A. tumefaciens, which are involved in determining virulence, as well as ipt, which is involved in cytokinin production (13, 16). In addition, ros mutants do not produce succinoglycan, and the repressive activity of Ros in Agrobacterium radiobacter is enhanced by Fe³⁺ and glucose in the culture medium (11, 14).

To elucidate the role of RosR in R. etli, we developed a genetic screen for genes transcriptionally regulated by RosR. RosR-regulated genes were identified by randomly inserting a reporter gene throughout the genome of a rosR mutant of R. etli and then comparing reporter gene expression in the presence and absence of rosR in trans. We identified the subset of those RosR-regulated genes involved in determining cell surface characteristics and nodulation competitiveness. This study represents the first broad-based screen of the entire genome to identify genes regulated by a member of this family of transcriptional regulators.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. Bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were grown in Luria-Bertani broth at 37°C, and R. etli strains were grown at 28°C in yeast extract mannitol (YEM) (35), tryptone-yeast extract (TY) (8), or Bergersen's synthetic medium supplemented with 1 mM methionine (BSM) (7). Solid media contained 1.5% agar, and antibiotics were used at the following concentrations: streptomycin, 200 µg/ml; spectinomycin, 100 µg/ml; nalidixic acid, 15 µg/ml; ampicillin, 50 µg/ml; tetracycline, 12 µg/ml; and kanamycin, 50 µg/ml. X-GlcA (5-bromo-4-chloro-3-indolyl glucuronic acid) was used at 50 µg/ml. Plasmid DNA was isolated from E. coli using the Qiaprep Kit (Qiagen Inc.). Restriction and modification enzymes were used according to the directions of the manufacturers (Promega Corp. and New England Biolabs). Plasmids were introduced into R. etli either by triparental mating using the helper plasmid pRK2013 or by biparental mating using E. coli strain S17-1 pir as the donor strain.

Identification of RosR-regulated insertions. R. etli strain CE3rosR was mutagenized with mTn5SSgusA40 by a previously described method (36), and transposon mutants were selected on BSM with appropriate antibiotics. Individual mutants were patched onto master plates in a grid of 48 mutants per plate (Fig. 1). These mutants were replicated onto TY without antibiotics, to avoid antibiotic carryover to the mating plates, using 48-prong metal replicators. E. coli S17-1 pir(pH8B1) was grown in Luria-Bertani broth overnight, and cells were pelleted, washed twice with TY to remove antibiotics, and resuspended in the original volume of TY. Two hundred microliters of the resuspended cells was plated onto TY plates to form the lawn of the donor strain. The R. etli mutants were replicated onto the lawn of S17-1 pir(pH8B1) and grown overnight at 28°C. These mating spots were then replicated onto BSM with appropriate antibiotics to select for the R. etli mutants carrying pH8B1. Both collections of strains (with and without pH8B1) were replicated into 96-well microtiter plates in which each well contained 150 µl of YEM with appropriate antibiotics. The microtiter plates were incubated at 28°C until both sets of plates contained dense cultures. The microtiter plates of the mutants with and without plasmid were replicated onto BSM containing X-GlcA and appropriate antibiotics. These plates were placed at 28°C, and corresponding colonies from each collection of strains were monitored for the appearance of a blue product over several days. Mutants that displayed different levels of accumulation of the blue color with and without pH8B1 were identified, the corresponding patch on the master plate was picked, and a single colony was isolated and retested for RosR regulation in quantitative enzyme assays.

View larger version (39K): [in this window] [in a new window]   FIG. 1.   Schematic diagram of the genetic screen used to identify RosR-regulated genes.

Quantitative GUS assays. Strains to be assayed were grown for 3 days in YEM broth. Fifty microliters of the culture was used to inoculate 2 ml of YEM broth and grown overnight. Enzyme assays were carried out on the overnight cultures by using 4-methylumbelliferyl--glucuronide as a substrate for the -glucuronidase (GUS) enzyme (22). Product accumulation was monitored using a TKO-100 fluorometer (Hoefer Scientific Instruments), and the bicinchoninic acid protein assay reagent (Pierce) was used to determine protein concentrations in calculating enzyme activities. Enzyme assays were repeated at least three times using independent cultures. Repression by RosR was calculated by dividing the average GUS activity of the mutant with pLAFR3 by the average GUS activity of the mutant with pH8B1. Activation by RosR was calculated by dividing the average activity of the mutant with pH8B1 by the average activity of the mutant with pLAFR3.

DNA sequencing of regions flanking the transposon insertions. Genomic DNA from each transposon mutant was cut with SacI and cloned into pBBR1MCS-3, selecting for the spectinomycin resistance gene carried on the transposon. The sequence of the flanking region was obtained by sequencing from the I end of the transposon with the primer 5'GGG AAT TCG GCC TAG GCG G3' and from the O end (the end with the promoterless gusA gene) with the primer 5'TTT CTA CAG GAC GTA ACA TAA GGG3'. The Big-dye cycle sequencing kit (Applied Biosystems, Inc.) was used, and the resulting reactions were analyzed at the University of WisconsinMadison Biotechnology Center. Sequences from both ends of the transposon were trimmed of transposon sequence and fused to obtain a single sequence of the insertion site. To avoid removal of any biologically relevant sequence information, we did not remove the sequences of small duplications at the site of the transposon insertion, which may have occurred as a result of the transposition event. The DNA sequences of the regions identified by more than one insertion were assembled into a single contig, which was used for sequence analysis. Sequence analysis was carried out in July 1999 using the BLASTn and BLASTx algorithms at the National Center for Biotechnology Information via the worldwide web (http://www.ncbi.nlm.nih.gov/) (1).

Screening of mutants for altered competitiveness. P. vulgaris cultivar Black Turtle seeds (Park Seed Co.) were surface disinfected by treatment with 95% ethanol for 30 s and with 1.6% sodium hypoclorite for 3 min and were planted in a sterilized sand-vermiculite (1.5:1) mixture. Bacterial strains were grown on TY plates with appropriate antibiotics, scraped from the plates, and resuspended in sterile water to an A₆₀₀ of 0.1 (approximately 10⁸ cells/ml). Either CE3003 or CE3013 was used as a kanamycin-resistant competitor strain. Inoculum mixtures were made by mixing the strains in 1:1 ratios. One milliliter of inoculum was applied to each planted bean seed. Serial dilutions of inocula were plated to determine cell numbers. Beans were placed in a growth chamber and watered with sterile nitrogen-free plant nutrient solution as needed for 21 days (2). Each treatment was applied to six plants, eight nodules were harvested from each plant, and bacterial strains in the nodules were identified by antibiotic resistance (6). Each nodulation competitiveness assay was repeated at least twice.

Nucleotide sequence accession numbers. The sequences from each end of the transposons in the transposon mutants were deposited in the GenBank database, and the accession numbers are indicated in Table 2.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of RosR-regulated genes. We developed a genetic screen to identify RosR-regulated genes (Fig. 1). A derivative of R. etli strain CE3 with the rosR gene deleted, CE3rosR, was mutagenized with a mini-Tn5 containing a promoterless gusA gene, which encodes GUS, at one end acting as a transcriptional reporter. rosR was introduced in trans into each of the mutants by a mass-mating technique. Individual transposon mutants were screened on indicator medium for GUS expression for differences in reporter gene expression with and without rosR in trans. Quantitative GUS enzyme assays confirmed the initial mutant phenotypes (Table 2).

Sequences of regions flanking transposon insertions. Analysis of the nucleotide sequences flanking the transposon insertions revealed that each mutant was the result of a unique insertion event. The insertions were in 43 different loci, and 7 loci were identified more than once. Most of the transposon insertions are oriented such that the gusA reporter gene is oriented in the same direction as the portion of the open reading frame (ORF) identified by sequence similarity. The exceptions are mutants MB009, MB041, MB050, and MB054. The RosR-dependent regulation in these mutants may be due to transcription from a RosR-regulated promoter downstream from and oriented convergently to the gusA gene. The transposon in MB039 is inserted in the opposite orientation but in the same site as that of MB041.

Identification of RosR-regulated genes that affect the cell surface. Mutants MB013 and MB065 produced colonies with altered morphology. Both mutants (without rosR in trans) produce colonies on YEM agar that initially appear to be similar to those of the rosR mutant, CE3rosR, but after 4 days of growth the colonies appear to be similar to those of the wild-type strain, CE3 (Fig. 2). When rosR is present in trans in these mutants, they produce colonies that are indistinguishable from the wild type at all times during growth (data not shown). The region flanking the insertion in MB013 has no similarity to any known genes in the database, and the region flanking the insertion in MB065 is similar to the exoY gene from Rhizobium sp. (Table 2).

View larger version (112K): [in this window] [in a new window]   FIG. 2.   Colony morphologies of mutants MB013 and MB065 in comparison to CE3 and CE3rosR. The same YEM plate is shown at 3 days (A) and 4 days (B) after inoculation.

Identification of RosR-regulated genes involved in nodulation or nodulation competitiveness. Each mutant strain with either pLAFR3 or pH8B1 in trans was singly inoculated onto beans to determine if each strain could nodulate beans. All mutant strains except MB015 nodulated beans with or without rosR supplied in trans. Mutant MB015 did not nodulate beans unless rosR was provided in trans. When MB015(pLAFR3) (a rosR mutant) was singly inoculated onto beans, either no nodules developed or only a few small white nodules developed on the bean roots. When MB015(pH8B1) (rosR⁺) was singly inoculated onto beans, normal-appearing nodules developed (data not shown).

View larger version (54K): [in this window] [in a new window]   FIG. 3.   Nodulation competitiveness of mutants MB013 and MB065. Each graph represents a separate treatment. The relative percentage of each strain in the inoculum is indicated by the bar on the left, and the relative percentage of nodules later occupied by each strain (or of nodules occupied by both strains [double occup.]) is indicated by the bar on the right. The treatments that demonstrate the increase in competitiveness of MB013 and MB065 are highlighted by heavy outlining. Standard errors are indicated adjacent to each column division.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

To understand the role of RosR in R. etli, we developed a genome-wide genetic screen to identify members of the RosR regulon. This screen was designed to identify genes that are negatively or positively regulated by RosR as well as genes directly or indirectly regulated by RosR. Based on the phenotypes of the rosR mutant, we expected to identify three classes of mutants: (i) mutants with an altered cell surface, (ii) mutants with altered nodulation competitiveness, and (iii) mutants with insertions in other RosR-regulated genes. All three classes were identified in the screen.

We identified 43 different RosR-regulated loci, and 7 of the loci were identified by more than one insertion. Two of the RosR-regulated genes affect both the cell surface and nodulation competitiveness of R. etli. Sixteen of the insertions are in regions with no significant sequence similarity to proteins in the sequence databases. Only one gene was positively regulated by RosR.

In the absence of rosR in trans, mutants MB013 and MB065 both produce colonies indistinguishable from the wild type after extended growth on solid media. The region flanking the insertion in MB013 shows no similarity to known genes, while the insertion in MB065 is in a region with similarity to exoY. ExoY is an essential part of the succinoglycan biosynthesis pathway, likely acting as a sugar transferase, and the RosR homolog, MucR, binds upstream of exoY in S. meliloti (9, 30). Assuming that the hydrophobic surface of the rosR mutant is due to derepression of one or more RosR-regulated genes, it is consistent that insertions in some RosR-regulated genes lead to a reversion to hydrophilic cell surfaces.

The two mutants with altered colony morphology were more competitive than the rosR mutant strain. Derepression of many RosR-regulated genes may lead to the great decrease in competitiveness observed in the rosR mutant; therefore, an insertion in any one of those genes would increase the competitiveness of a rosR mutant. The correlation between the subset of RosR-regulated genes that affect the cell surface properties and the subset that affect nodulation competitiveness suggests that the altered competitiveness of the rosR mutant is due to the dramatic changes in cell surface characteristics.

Mutant MB015 without rosR in trans nodulated poorly, yet when rosR was present in trans, the mutant nodulated normally and was unaffected in nodulation competitiveness. The insertion in MB015 is in a region with similarity to exsH from S. meliloti. ExsH is an endoglycanase that cleaves high-molecular-weight EPS into lower-molecular-weight forms (37). In S. meliloti, low-molecular-weight forms of EPS (either EPS I or EPS II) are important for establishing a successful symbiosis, probably acting as a signal molecule to the plant host (4, 20). Interestingly, S. meliloti mucR mutants do not produce low-molecular-weight EPS II, and when mucR mutants are blocked in EPS I production, they do not nodulate the plant host (20). If a low-molecular-weight polysaccharide signal is a common theme in all rhizobial interactions, it is possible that this gene is needed in R. etli to produce such a signal molecule in the rosR mutant background. Further studies need to confirm that it is indeed the gene with similarity to exsH that is responsible for this phenotype and that it is not due to polar effects from the insertion on other downstream genes.

We identified many other RosR-regulated genes that did not affect colony morphology or nodulation competitiveness, and sequence analysis of the regions flanking the insertions suggests hypotheses about the functions of some of these RosR-regulated genes. We identified genes that may be involved in polysaccharide and carbohydrate metabolism, genes that may be involved in survival in the rhizosphere, and genes similar to those that are regulated by Ros in A. tumefaciens (Table 2).

We identified genes that encode proteins with similarity to ExoB, ExoY, ExsH, PrsD, PssK, and PlyA, which are involved in EPS synthesis in other rhizobial species (12, 19, 30, 37). PrsD is a component of a secretion system involved in the export of the ExsH and PlyA proteins in Rhizobium meliloti and Rhizobium leguminosarum, respectively (19, 38). Mutant MB002 has an insertion in a region that encodes a protein with similarity to a sugar acetylase. Other RosR-regulated genes encoding proteins with similarity to glucose dehydrogenase, trehalose-phosphate synthase, and a transcriptional repressor of the sucrose degradation operon may also be involved in carbohydrate metabolism. Altered expression of these genes in rosR mutants is consistent with the activities of Ros in A. tumefaciens and MucR in R. meliloti, both of which affect EPS production.

Other mutants have insertions in genes that may affect bacterial fitness in the rhizosphere. Cellulose synthesis is important for A. tumefaciens attachment to its plant host (27), and mutants MB025 and MB037 contain insertions in a region with similarity to cellulose synthase genes, while the insertion in MB038 is in a region with similarity to celR2, a positive regulator of the genes involved in cellulose synthesis in R. leguminosarum bv. trifolii (3). MB042 has an insertion just upstream of a region with similarity to an ORF in the picA locus of A. tumefaciens, which is involved in altering the cell surface characteristics of the bacterium in response to compounds in plant cell extracts (31). Mutants MB018 and MB019 have insertions adjacent to regions with similarity to genes involved in the synthesis and transport of opines, carbon and nitrogen sources whose production is symbiotically regulated in S. meliloti (28).

Our identification of RosR-regulated genes suggests that the role of RosR in R. etli is similar to that of Ros in A. tumefaciens. Like Ros, RosR is involved in regulation of a region with similarity to vir genes and may affect cytokinin and heme synthesis (11, 13, 16). The predicted product of the ORF upstream from and oriented divergently to the gusA reporter gene in MB043 has a high level of conservation with the VirC1 sequence, containing 87% amino acid identity over 62 amino acids, while the ORF downstream from and in the same orientation as the gusA gene encodes a predicted protein that is 44% identical over 47 amino acids to VirD3. The insertion in MB006 is within an ORF likely to encode a coproporphyrinogen III oxidase, which is involved in the formation of heme (23). MB011 has the transposon insertion in a region with similarity to a gene that may be involved in cytokinin production and virulence in the plant pathogen Rhodococcus fascians, and rhizobial production of cytokinin is implicated in Rhizobium-induced leaf curl syndrome of pigeon pea (17, 34).

Our dissection of the RosR regulon has revealed that RosR affects expression of many functionally diverse genes in R. etli, and it seems likely that the dominant role of RosR is to influence gene expression by transcriptional repression. Surprisingly, the role of RosR shares features of that of Ros in A. tumefaciens, regulating homologs of vir genes as well as affecting polysaccharide production. The presence of vir homologs in a rhizobial species has not been previously reported, and preliminary sequence obtained from pMB043 indicates that homologs of other vir genes are also present on pMB043 (M. A. Bittinger, unpublished data).

As more bacterial genomes are sequenced, there is an increasing need to link an understanding of gene expression with sequence information. DNA sequence alone can be used to predict gene function, but delineation of pathways of coordinate regulation requires functional genetic approaches. Genetic approaches such as we employed in this study allow us to begin to dissect complex regulatory pathways that will complement the anticipated full genome sequence of a rhizobial species.