NolR Regulates Diverse Symbiotic Signals of Sinorhizobium fredii HH103

676 / Molecular Plant-Microbe Interactions MPMI Vol. 17, No. 6, 2004, pp. 676–685 Publication no. M-2004-0405-01R. © 2004 The American Phytopathological Society NolR Regulates Diverse Symbiotic Signals of Sinorhizobium fredii HH103 José María Vinardell,1 Francisco Javier Ollero,1 Ángeles Hidalgo,1 Francisco Javier López-Baena,1 Carlos Medina,1 Kalojan Ivanov-Vangelov,1 Maribel Parada,1 Nuria Madinabeitia,1 María del Rosario Espuny,1 Ramón Andrés Bellogín,1 María Camacho,2 Dulce-Nombre Rodríguez-Navarro,2 María Eugenia Soria-Díaz,3 Antonio M. Gil-Serrano,3 José Enrique Ruiz-Sainz1 1Departamento de Microbiología, Facultad de Biología, Universidad de Sevilla, Avda. Reina Mercedes 6, C.P. 41012, Sevilla, Spain; 2CIDA “Las Torres y Tomejil”, Apdo. Oficial, Alcalá del Río, Sevilla, Spain; 3Departamento de Química Orgánica, Facultad de Química, Universidad de Sevilla, Apdo. 553, 41071-Sevilla, Spain Submitted 3 July 2003. Accepted 2 February 2004. We have investigated in Sinorhizobium fredii HH103-1 (=HH103 Strr) the influence of the nolR gene on the production of three different bacterial symbiotic signals: Nod factors, signal responsive (SR) proteins, and exopolysaccharide (EPS). The presence of multiple copies of nolR (in plasmid pMUS675) repressed the transcription of all the flavonoid-inducible genes analyzed: nodA, nodD1, nolO, nolX, noeL, rhcJ, hesB, and y4pF. Inactivation of nolR (mutant SVQ517) or its overexpression (presence of pMUS675) altered the amount of Nod factors detected. Mutant SVQ517 produced Nod factors carrying N-methyl residues at the nonreducing N-acetyl-glucosamine, which never have been detected in S. fredii HH103. Plasmid pMUS675 increased the amounts of EPS produced by HH103-1 and SVQ517. The flavonoid genistein repressed EPS production of HH103-1 and SVQ517 but the presence of pMUS675 reduced this repression. The presence of plasmid pMUS675 clearly decreased the secretion of SR proteins. Inactivation, or overexpression, of nolR decreased the capacity of HH103 to nodulate Glycine max. However, HH103-1 and SVQ517 carrying plasmid pMUS675 showed enhanced nodulation capacity with Vigna unguiculata. The nolR gene was positively identified in all S. fredii strains investigated, S. xinjiangense CCBAU110, and S. saheli USDA4102. Apparently, S. teranga USDA4101 does not contain this gene. Additional keywords: LCO, symbiotic interaction. Rhizobia are α-proteobacteria belonging to the genera Rhizobium, Sinorhizobium, Bradyrhizobium, Mesorhizobium, Allorhizobium, and Azorhizobium. These soil bacteria establish specific symbiotic relationships with leguminous plants resulting in the formation of nitrogen-fixing nodules (Cullimore et al. 2001; Geurts and Bisseling 2002; Perret et al. 2000). The nodulation process is a complex event that requires a molecular dialogue between the partners. In response to flavonoids exuded by the roots, rhizobia secrete specific signal molecules called Nod factors or lipochitooligosaccharides (LCO) that induce in the plant the development of root nodules. Putative receptors for the rhizobial Nod factors recently have been identified in several leguminous plants (Limpens et al. 2003; Madsen et al. 2003; Spaink 2002). Rhizobial nodulation genes (nod, nol, and noe) are involved in Nod factor production and constitute the nod regulon, whose expression is finely regulated by both positive and negative elements (Schlaman et al. 1998). In the presence of flavonoid inducers, the transcription of most of the nodulation genes is activated by the constitutively expressed NodD protein. The NodD protein belongs to the bacterial LysR regulatory family, and it binds to conserved nucleotidic sequences, called nod boxes, which are present in the promoter regions of the nod operons. On the other side, some Sinorhizobium meliloti strains produce a 13-kDa protein, NolR, that represses the expression of the nodulation genes, and it is necessary for optimal nodulation on alfalfa (Kondorosi et al. 1991). DNA hybridization experiments have shown that nolR homologous sequences are present in species belonging to the Rhizobium and Sinorhizobium genera, but not in species from Mesorhizobium, Bradyrhizobium, Azorhizobium, and Agrobacterium genera (Kiss et al. 1998). The S. meliloti NolR protein binds to a specific DNA target sequence that is present in the nolR promoter as well as in the promoter regions of nod genes involved in the synthesis of the Nod factor core (Cren et al. 1995). However, nod operons required for the decoration of the Nod factors are not repressed by NolR. Because of this, it has been suggested that the activity of NolR results in the preferential synthesis of fully decorated S. meliloti Nod factors that, in low amounts, appear to be optimal for nodulation. NolR appears to belong to the AsrR regulatory family, a group of proteins that comprises both positive and negative regulators (Kiss et al. 1998). Proteomic analyses of S. meliloti have revealed that NolR is a global regulatory protein that increases, or decreases, the levels of numerous proteins participating in different cellular processes (Chen et al. 2000). In addition to Nod factors, other rhizobial molecules are required for an efficient nodulation process (Fraysse et al. 2003; Perret et al. 2000). Thus, several rhizobia secrete symbiotically active proteins. Rhizobium leguminosarum bv. viciae secretes a Ca2+-binding protein, NodO, that probably forms cation-specific channels in cell membranes of leguminous plants and that allows nodulation of Vicia hirsuta by nodCorresponding author: José M. Vinardell; Telephone: +34-954557121; Fax: +34-954557830; E-mail: [email protected] Nucleotide and amino acid sequence data are available in the EMBL database under accession numbers AY186253, AY194594, AY194595, AY194596, AY194597, AY194598, AY194599, and AY194600. Vol. 17, No. 6, 2004 / 677 FELMNT mutants of this bacterium (Economou et al. 1990). By a type-III secretion system (TTSS), Sinorhizobium sp. NGR234 and S. fredii strains secrete several signal responsive (SR) proteins in response to inducing flavonoids in a NodD1dependent manner (Krishnan et al. 1995; Viprey et al. 1998). In S. fredii USDA257, the locus nolXWBTUV, involved in this secretion process, regulates the nodulation of Glycine max in a cultivar-specific manner (Meinhardt et al. 1993). This locus also determines S. fredii nodulation capacity with other legumes, such as Erythrina spp. (Bellato et al. 1997). Genes encoding a TTSS recently have been found in some other rhizobia, such as Bradyrhizobium japonicum (Krause et al. 2002). Rhizobial exopolysaccharides (EPSs), lipopolysaccharides (LPSs), K-antigen capsular polysaccharides (KPSs), and the cyclic glucans also appear to play important roles in different stages of the nodulation process (Fraysse et al. 2003). In this study, we have isolated and studied the nolR gene of S. fredii HH103. Our results indicate that S. fredii HH103 NolR regulates not only the production of Nod factors but also other rhizobial symbiotic signals, such as EPS and SR proteins. The absence of NolR, or its overexpression, also affects the symbiotic interaction of S. fredii HH103 with G. max cv. Williams and Vigna unguiculata in a host-plant-dependent manner. In addition, we show that the nolR nucleotidic sequence is well conserved among several Sinorhizobium spp. and in Rhizobium sp. NGR234. RESULTS Isolation of the S. fredii nolR gene. Two primers, nolRint-f and nolRint-r, were designed from the homologous regions of the nolR coding sequences of S. meliloti AK631 (X59050) and R. leguminosarum bv. viciae TOM (AJ001934), and used for polymerase chain reaction (PCR) amplification of a 227-bp fragment of S. fredii HH103. DNA sequencing of the HH103-amplified fragment showed a high degree of identity to the nolR gene of AK631 (89%); therefore, it was used as a hybridization probe to isolate cosmid pMUS671 from a HH103 genomic library. A 2,547-bp EcoRI fragment from cosmid pMUS671 containing the S. fredii HH103 nolR gene was subcloned into plasmid pBluescript (generating plasmid pMUS672) and sequenced (accession number AY186253). This was the unique EcoRI fragment present in the HH103-1 genomic DNA showing hybridization to the probe (data not shown), indicating that there is only one copy of the nolR gene in the genome of S. fredii HH103-1. The 2,547-bp EcoRI fragment contains two complete and one partial open reading frame (ORF) showing the same polarity. The nolR gene extends between positions 307 and 663, encoding for a 118-amino acid (aa) polypeptide that shows 89 and 75% of identity to the NolR proteins of S. meliloti AK631 and R. leguminosarum bv. viciae TOM, respectively. The NolR-binding site (atgcatcacgGctaat, nucleotide positions 897 to 912 of X59050) that appears upstream of the S. meliloti nolR gene is also present with only one mismatch in S. fredii HH103 (atgcatcacgTctaat, nucleotide positions 247 to 262 of AY186253). The second ORF, called ORF2, is located between positions 795 and 2,195 and encodes for a putative polypeptide of 466 aa that is homologous to a putative omega amino acid pyruvate aminotransferase of S. meliloti 1021 (97% of identity, AL591790.1) and to a beta-alanine-pyruvate transferase of Mesorhizobium loti (83% of identity, AP002997.2). The HH103 sequenced fragment also contains the 5′ terminus of a putative ORF (starting at position 2,284) whose encoded product (87 aa) shows 94 and 50% of identity with a putative adenine deaminase of S. meliloti (AL591790.1) and with an adenine deaminase of Bacillus subtilis (X83795), respectively. The nucleotide homology (80.3% of identity) of the S. fredii fragment to that of S. meliloti AK631 extends between positions 1 and 744 in the former and 651 and 1,402 in the latter. In addition, the HH103-sequenced fragment shows homology (88% of identity) to several DNA segments located between nucleotides 212,557 and 215,482 of a chromosomal region of S. meliloti 1021 (AL591790.1). This region contains the nonfunctional nolR gene of this strain (Cren et al. 1994), as well as genes that putatively encode for an omega amino acid pyruvate aminotransferase and an adenine deaminase mentioned above. S. fredii NolR represses nod genes. The 2,547-bp EcoRI fragment containing the HH103 nolR gene was subcloned into plasmid pMP92 (generating plasmid pMUS675) and transferred to different S. fredii HH103 mutants carrying a Tn5-lacZ insertion into different flavonoidinducible genes that are responsible for the production of Nod factors (nodA, nolO, noeL, and nodD1), for the secretion of SR proteins (nolX, rhcJ, and rhcQ), or for other nondetermined functions (hesB and a locus homologous to the Rhizobium sp. NGR234 y4pF). Plasmid pMUS675 decreased the level of β-galactosidase activity exhibited by all the HH103 Tn5-lacZ derivatives assayed in the presence of the flavonoid genistein (Table 1). In contrast, transfer of plasmid pMUS810 (similar to pMUS675 but carrying the Ω interposon inserted into the nolR gene) to S. fredii SVQ116 (HH103 Rifr nodA::Tn5-lacZ) had no effect on the bacterial β-galactosidase activity. Plasmid pMUS675 also was transferred to strain SVQ296, a derivative of HH103-1 carrying a Tn5-lacZ insertion elsewhere in the bacterial genome, in which its β-galactosidase activity is not influenced by the exogenous addition of flavonoids. Plasmid pMUS675 had no effect on the level of βgalactosidase activity of strain SVQ296 in either the presence or absence of genistein. These results indicate that the capacity of plasmid pMUS675 to repress the expression of nodulation genes and other genes showing transcriptional activation by flavonoids is provided by the HH103 nolR gene. Computer analysis of the S. fredii sequences available in the databases revealed the presence of putative NolR-binding sites in the upstream regions of several genes whose expression has been shown to be influenced by this protein, such as nodD1, hesB, and the nodABCIJnolOnoeI and nodZnoeLnolK operons (Fig. 1). In addition, the intergenic region between the divergently transcribed nodD2 and ttsI genes contains a nod box (in the direction of ttsI) and a well-conserved NolR-binding site. The ttsI-encoded product appears to act as a transcriptional activator of genes involved in the TTSS of Rhizobium sp. NGR234 and B. japonicum (Krause et al. 2002; Viprey et al. 1998). In S. fredii HH103, the putative NolR-binding sites of the nodD1, nodD2, and nolR genes are located in the complementary strand, as has been described for the nolR, nodD1, nodD2, and nodD3 genes of S. meliloti (Cren et al. 1995). Mutation of the S. fredii HH103 nolR gene alters the production of Nod factors. An nolR mutant derivative (SVQ517) of strain HH103 was generated (discussed below). In the absence of flavonoids, the β-galactosidase activity of HH103 carrying plasmid pMP240 (it contains a transcriptional fusion between the R. leguminosarum bv. viciae nodA promoter and the lacZ gene) was 57% of that shown by its nolR derivative SVQ517 carrying this plasmid (550 ± 31 and 960 ± 69 Miller units, respectively). In the presence of genistein, β-galactosidase activities of HH103 (pMP240) and SVQ517 (pMP240) were 11,071 ± 678 and 13,689 ± 1,506, respectively. 678 / Molecular Plant-Microbe Interactions S. fredii strains HH103-1, SVQ517, and their respective derivatives carrying plasmid pMUS675 were cultured in the presence and absence of the S. fredii nod-inducer flavonoid genistein. Nod factors isolated from these bacterial cultures were analyzed by reverse-phase thin-layer chromatography (RPTLC) (data not shown). In the absence of flavonoids, none of the tested strains produced detectable amounts of Nod factors. When genistein was present, subtle differences were observable. Nod factors production by the HH103-1 nolR mutant appeared to be slightly higher than that of its parental strain HH103-1. The presence of plasmid pMUS675 did not show any detectable effect on the amount of Nod factors produced by HH103-1, but it slightly reduced the level of LCO secreted by the nolR mutant. Experiments using high-performance liquid chromatography (HPLC) also indicated that nolR inactivation increased Nod factor production. Strains HH103-1 and SVQ517 secreted approximately 3.9 and 17.7 mg of LCO, respectively, when they were grown in 3 liters of B– medium (Spaink et al. 1992) supplemented with genistein. The presence of plasmid pMUS675 did not significantly affect the level of Nod factors secreted by HH103-1 (approximately 4.6 mg) but clearly reduced the amount of Nod factors produced by SVQ517 (approximately 6.8 mg). Fast-atom-bombardment mass spectrometry (FABMS) analyses revealed differences on the set of LCO produced by strains HH103-1 and SVQ517 (Table 2). Twenty-two different LCO could be identified in SVQ517. Some of them, such as those carrying N-methyl residues at the nonreducing Nacetyl-glucosamine, have never been found before in the parental strain S. fredii HH103-1. The presence of plasmid pMUS675 in SVQ517 abolished the production of these Nmethylated Nod factors. Overexpression of nolR also exerted an effect on the specific set of Nod factors detected, increasing the amount of fucosylated LCO with respect to the methyl-fucosylated ones. These results indicate that the nolR gene not only affects the total amount of LCO secreted by S. fredii HH103 but also the relative quantities of the different molecular species produced. S. fredii HH103 nolR influences SR proteins secretion and EPS production. The nolR gene affects the expression level of several HH103 genes involved in the TTSS (such as nolX, rhcQ, or rhcJ); Fig. 1. Putative NolR-binding sites in the promoter regions of different Sinorhizobium fredii nod genes. For each sequence, nucleotides matchin g the consensus sequence are underlined. Table 1. β-Galactosidase activity of HH103-1 derivatives carrying lacZ insertions into different flavonoid-inducible g enes in the presence or absence o f plasmid pMUS675 β-Galactosidase activity (Miller units) Strain Gene carrying lacZ – Genistein + Genistein Expression (%)z SVQ116 nodA 117 ± 20 887 ± 189 SVQ116(pMUS675) 117 ± 20 413 ± 58 47 SVQ116(pMUS810) 107 ± 7 871 ± 127 98 SVQ118 nolX 162 ± 5 678 ± 81 … SVQ118(pMUS675) 175 ± 13 237 ± 12 35 SVQ120 rhcQ 38 ± 13 135 ± 24 … SVQ120(pMUS675) 47 ± 20 56 ± 11 43 SVQ121 nolO 54 ± 10 212 ± 36 … SVQ121(pMUS675) 52 ± 12 124 ± 26 58 SVQ123 Homologous to y4pF of Rhizobium sp. NGR234 117 ± 0 870 ± 59 … SVQ123(pMUS675) 145 ± 5 138 ± 30 16 SVQ124 hesB 179 ± 35 509 ± 52 … SVQ124(pMUS675) 158 ± 42 174 ± 3 34 SVQ287 noeL 64 ± 16 286 ± 16 … SVQ287(pMUS675) 73 ± 1 155 ± 1 54 SVQ288 rhcJ 247 ± 60 1,255 ± 260 … SVQ288(pMUS675) 298 ± 67 524 ± 167 42 SVQ502 nodD1 99 ± 6 109 ± 4 … SVQ502(pMUS675) 42 ± 10 46 ± 7 42 SVQ296 Not characterized (not flavonoid-inducible) 277 ± 14 292 ± 35 … SVQ296(pMUS675) 299 ± 43 276 ± 31 95 z Numbers refer to the percentage of β-galactosidase activity of genistein-induced cultures of each mutant carr y in g plasmid pMUS675, or pMUS810, compared with that observed in the absence of this plasmid. Vol. 17, No. 6, 2004 / 679 therefore, we investigated the effect of nolR inactivation or nolR overexpression on the secretion of SR proteins. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDSPAGE) experiments showed that S. fredii HH103-1 secretes at least five proteins in response to the presence of genistein (Lyra 2001). Two of these proteins show an apparent molecular weight that is lower than 6.5 kDa. The others are of approximately 60, 40, and 20 kDa (Fig. 2, lane 2, bands marked with black asterisks). Inactivation of the S. fredii HH103 nolR gene did not produce any clear alteration on the SR-protein profile. However, the presence of pMUS675 in strains HH103-1 and SVQ517 exerted a dramatically negative effect on the secretion of SR proteins because only small amounts of the two largest proteins were detected in the supernatants of bacterial cultures grown in the presence of genistein (Fig. 2, lanes 7 and 9). The presence of the empty vector plasmid pMP92 or plasmid pMUS810 (nolR::Ω) affected the amounts of SR proteins secreted by neither HH103-1 nor SVQ517 (data not shown). These results indicated that, in S. fredii HH103, nolR also represses the secretion of SR proteins. The procedure to isolate extracellular proteins includes one precipitation step, in which acetone is added to cultures supernatants. This treatment with acetone also provokes the aggregation of the EPS that is present in the medium. In the course of the isolation of the SR proteins, it was serendipitously found that the amount of EPS recovered from S. fredii HH103 cultures was influenced by the absence, or the overexpression, of the nolR gene. In the absence of genistein, the amount of EPS recovered (7.7 mg) from SVQ517 cultures was approximately nine times lower than that (69.0 mg) from its parental strain HH103-1. Although plasmid pMUS675 increased the amount of EPS recovered (93.8 mg) from HH103-1 cultures, this positive effect on bacterial polysaccharide production was even clearer in SVQ517 (98.5 mg). The presence of genistein reduced the amount of EPS recovered from all of the bacterial cultures investigated, although this reduction was attenuated when bacteria carried plasmid pMUS675 (10.4 and 69.4 mg for HH1031 and HH103-1 pMUS675, respectively). Moreover, in the presence of this nod inducer, EPS was recovered from SVQ517 cultures only if this mutant also was carrying plasmid pMUS675 (42.3 mg). The presence of the empty vector pMP92 did not affect the amount of EPS recovered from Table 2. Summary of the lipochitooligosaccharide (LCO) structures identified as being produced by Sinorhizobium fredii strains HH103-1 and SVQ517, with and without plasmid pMUS675, in the presence of genistein (3.7 µM)x Strain [M+Na]+ m/z LCO structure HH103-1 HH103-1 (pMUS675) SVQ517 SVQ517 (pMUS675) 990 III(C16:1, Fuc) – – + – 1004 III(C16:1, MeFuc) –y – + – 1018 III(C18:1, Fuc) – – + – 1032 III(C18:1, MeFuc) –y – + – 1046 III(C18:1, NMe,MeFuc) – – + – 1048 III(C18:0, NMe,MeFuc) – – + – 1193 IV(C16:1, Fuc) –y + z + +z 1195 IV(C16:0, Fuc) – +z – +z 1207 IV(C16:1, MeFuc) –y – + + 1209 IV(C16:0, MeFuc) + + + + 1221 IV(C18:1, Fuc) + +z + +z 1223 IV(C18:0, Fuc) – +z – +z 1235 IV(C18:1, MeFuc) + + + + 1237 IV(C18:0, MeFuc) + + + + 1249 IV(C18:1, NMe,MeFuc) – – + – 1251 IV(C18:0, NMe,MeFuc) – – + – 1396 V(C16:1, Fuc) –y – + + 1398 V(C16:0, Fuc) –y + + + 1410 V(C16:1, MeFuc) + + + + 1412 V(C16:0, MeFuc) + + + + 1424 V(C18:1, Fuc) + +z – + 1438 V(C18:1, MeFuc) + + + + 1440 V(C18:0, MeFuc) + + + + 1452 V(C18:1, NMe,MeFuc) – – + – 1454 V(C18:0, NMe,MeFuc) – – + – x LCOs were extracted from 3 liters of culture. y These LCOs have been detected previously when 10 liters of culture were used (Gil-Serrano et al. 1997). z The molecular ratio of these fucosylated LCOs with respect to their methyl-fucosylated homologues is higher than in HH103-1, accordin g to the intensit y of the pseudomolecular ion in mass spectrometry analyses. Fig. 2. Effect of NolR on the secretion of si g nal responsive (SR) proteins b y Sinorhizobium fredii HH103. Lanes 1 and 2: HH103-1; lanes 3 and 4: SVQ517; lane 5: molecular weight marker (sodium dodecyl sulfate polyacrylamide gel electrophoresis standard broad range from Bio-Rad, CA); lanes 6 and 7: HH103-1(pMUS675); lanes 8 and 9: SVQ517(pMUS675). Lanes 2, 4, 7, and 9: extracellular proteins from cultures g rown in the presence of genistein 3.7 µ M. Molecular wei g hts of the marker are indicate d on the left. SR proteins of S. fredii HH103 are indicated with an asterisk. 680 / Molecular Plant-Microbe Interactions HH103-1 cultures (66.9 and 13.3 mg in the absence or presence of genistein, respectively). Similar effects of the nolR gene on EPS production were observed when bacteria were grown on solid B– medium (data not shown). In the absence of genistein, strain HH103-1 was more mucous than SVQ517. However, conjugal transfer of plasmid pMUS675 to HH103-1 and SVQ517 produced transconjugants that were equally mucous. On the other hand, in the presence of genistein, the production of EPS was evident only when plasmid pMUS675 was present. Neither the presence of the empty vector pMP92 in HH103-1 and SVQ517 nor that of plasmid pMUS810 (nolR::Ω) in HH103-1 affected the amount of EPS produced. The nolR gene increases the amount of EPS produced by S. fredii HH103; therefore, we investigated whether the absence, or overexpression, of nolR produces any alteration of the bacterial LPS profile in SDS-polyacrylamide gels. The LPS profiles of strains HH103-1, SVQ517, and their derivatives carrying plasmid pMUS675 apparently were identical (data not shown). NolR affects the symbiotic behavior of S. fredii HH103. Plant tests were carried out to investigate the role of the nolR gene on the symbiotic interaction of S. fredii HH103 with two host plants, G. max cv. Williams and V. unguiculata (cowpea). Three different parameters were analyzed: plant-top dry weight, nodule number, and dry weight of nodules. Number and dry weight of nodules formed by G. max plants inoculated with mutant SVQ517 were significantly lower than those formed by soybean plants inoculated with the parental strain HH103-1 (Table 3). This reduction in the nodulation capacity exhibited by SVQ517 resulted in a significant decrease of the plant-top dry weight compared with that of soybean plants inoculated with HH103-1. The presence of plasmid pMUS675 in mutant SVQ517 slightly increased all the parameters analyzed, although significant differences were not scored. Surprisingly, the presence of plasmid pMUS675 in strain HH103-1 reduced its capacity to nodulate soybeans. These results indicate that not only the absence of a functional nolR gene, but also its overexpression, is detrimental for nodulation of S. fredii HH103 on soybean Williams. Additional plant tests showed that the nodulation rate of soybean plants inoculated with SVQ517 was significantly delayed with respect to its parental strain HH103-1. This is reflected by the mean number of nodules per plant for HH1031 and SVQ517: 0.6 versus 0.5 at day 10, 5.1 versus 1.0 at day 12, 11.8 versus 2.3 at day 17, 13.1 versus 7.1 at day 21, and 20.0 versus 11.9 at day 25. The number of nodules formed by HH103-1 and SVQ517 became significantly different 12 days after inoculation (at the significance level of α = 5%, MannWhitney nonparametric test). Competition experiments between strains HH103-1 and SVQ517 also were carried out with soybean Williams. When competitors were mixed at a 1:1 inoculate ratio (3 × 108 cells per competitor), nodule occupancy by SVQ517 was only 19.3%. Plant tests with V. unguiculata showed different results (Table 3). Number and dry weight of nodules as well as planttop dry weight of cowpea plants inoculated with SVQ517 were not significantly different (α = 5%) from those scored by inoculation with the parental strain HH103-1. The presence of plasmid pMUS675 in both strains produced a general increase in the three parameters analyzed. Plant-top dry weight of cowpea plants inoculated with SVQ517(pMUS675) was significantly higher (α =5%) than that of plants inoculated with SVQ517. The presence of pMUS675 in strain HH103-1 also significantly increased the number of nodules formed by cowpea plants. NolR is well conserved among Sinorhizobium and Rhizobium spp. In order to investigate the presence of the nolR gene in other S. fredii strains and related rhizobia, we designed several pairs of primers from the sequence of the nolR region of S. fredii HH103-1 that allow amplification of either the complete nolR coding sequence or an internal fragment of this gene. By using the pair of primers nolRupst/nolRdwst (discussed below), we could amplify a 479-bp fragment containing the complete nolR gene from S. fredii strains USDA205, USDA257, SMH12, and 042B(s) (accession numbers AY194594, AY194595, AY194596, and AY194597, respectively). The primer pair nolRupst/nolR-r led us to amplify a fragment of 367 bp containing the nolR gene from Rhizobium sp. NGR234 (AY194598). None of these primers amplified nolR-homologous sequences from S. xinjiangense CCBAU110, S. saheli USDA4102, or S. teranga USDA4101. However, primers nolRint-f and nolRint-r allowed the amplification of an Table 3. Plant responses to inoculation of Glycine max cv. Williams and Vigna unguiculata with Sinorhizobium fredii strains HH103-1, SVQ517, and thei r derivatives carrying plasmid pMUS675v Legume tested, inoculantw Plant-top dry weight (g)x Number of nodules Nodule dry weight (mg) G. maxy HH103-1 5.04 ± 0.88 a 268.4 ± 23.4 a 501.5 ± 44.3 a SVQ517 2.98 ± 0.68 b 115.0 ± 32.9 b 258.0 ± 69.3 b HH103-1(pMUS675) 2.93 ± 0.89 b 161.8 ± 63.5 ab 264.8 ± 75.0 b SVQ517(pMUS675) 3.11 ± 0.82 b 158.3 ± 37.3 b 297.5 ± 52.3 b V. unguiculataz HH103-1 2.49 ± 0.88 ab 234.3 ± 62.9 a 258.3 ± 75.2 a SVQ517 1.88 ± 0.59 a 282.6 ± 110.0 ab 255.6 ± 85.4 ab HH103-1(pMUS675) 3.45 ± 1.21 ab 593.0 ± 149.0 b 403.4 ± 136.0 ab SVQ517(pMUS675) 3.76 ± 0.88 b 401.0 ± 186.2 ab 451.3 ± 103.7 b v Data represent averages of seven or eight plants for soybean and of five or six plants for cowpea. w Bacteria isolated from 12 nodules formed by each inoculant showed the expected resistance markers. x Plan-top dry weight of noninoculated plants was 0.45 ± 0.25 g for soybean and 0.12 ± 0.02 g for cowpea. y For soybean plants, numbers in the same column flanked by the same letter are not significantly different at the significance level α = 5% followin g the nonparametric test of Kruskal-Wallis. The numbers of nodules formed by HH103-1 and HH103-1(pMUS675) were significantly different at α = 10% (and at the α = 1% following the nonparametric test of Mann-Whitney). z For cowpea plants, numbers in the same column flanked by the same letter are not significantly different at the significance level α = 5% followin g the nonparametric test of Kruskal-Wallis. The dry weight of nodules formed by SVQ517 and SVQ517(pMUS675) were significantly different at α = 10%. Pairwise comparisons using the nonparametric test of Mann-Whitney also allowed the detection of other significant differences. The number of nodules formed by SVQ517 was significantly lower (α = 2%) than that formed by HH103-1(pMUS675). Plant-top dr y wei g ht of cowpea plants inoculated with SVQ517 was significantly lower (α = 2%) than that of plants inoculated with HH103-1(pMUS675). Nodule dry wei g ht of plants inoculated with SVQ517 was significantly lower (α = 3%) than that of plants inoculated with SVQ517(pMUS675). Vol. 17, No. 6, 2004 / 681 internal fragment (188 bp) of the nolR gene of S. xinjiangense CCBAU110 (AY194599) and S. saheli USDA4102 (AY194600). Sequence analysis of the different PCR products obtained revealed that i) the nolR gene of all S. fredii strains and Rhizobium sp. NGR234 encodes a deduced polypeptide of 118 aa and ii) the nolR locus is well conserved among the different (sino)rhizobia included in this study. In fact, the NolRdeduced sequences of S. fredii strains HH103, USDA205, SMH12, and 042B(s) were identical. The partial polypeptide sequences (62 residues) deduced from the internal nolR sequences of S. xinjiangense CCBAU110 and S. saheli USDA4102 also were 100% identical to the corresponding region of the S. fredii HH103-1 NolR protein. The NolRdeduced sequences from S. fredii USDA257 and Rhizobium sp. NGR234 were 98.3% identical, and the same percentage of identity was found between each of these sequences and that of the S. fredii HH103 NolR polypeptide. DISCUSSION Participation of the bacterial NodD transcriptional activator in conjunction with inducer flavonoids is required for nod gene expression in most rhizobia. In addition to nodD, which can be present in one or in several copies, there are others positive regulatory elements of nod gene expression, such as syrM in S. meliloti and Rhizobium sp. NGR234 (Schlaman et al. 1998). Several repressors of nod genes also have been reported. Fellay and associates (1998) have reported that NodD2 of Rhizobium sp. NGR234 is involved in the repression of the nodABC operon. In addition, the presence of Nod factors in cultures of B. japonicum USDA110 induces the expression of nolA, which in turns activates nodD2 and, thus, the transcription of nodulation genes is repressed (Loh and Stacey 2001). The nolR gene initially was found in S meliloti AK631 (Kondorosi et al. 1991), in which its relevance for an optimal nodulation on alfalfa was demonstrated. Some S. meliloti strains, such as 1021, do not contain a functional nolR gene (Cren et al. 1994). More recently, Kiss and associates (1998) reported the presence of a functional nolR gene in R. leguminosarum bv. viciae strain TOM and also demonstrated by hybridization the presence of nolR homologous sequences in several Rhizobium and Sinorhizobium spp., including S. fredii and Rhizobium sp. NGR234. By using PCR and hybridization techniques, Fellay and associates (1998) demonstrated that NGR234 harbors a single, chromosomal copy of nolR. In this article, we report the isolation and sequence of the nolR gene of five different S. fredii strains and also provide the sequence of the nolR gene of Rhizobium sp. NGR234. Internal fragments of the nolR genes of S. xinjiangense and S. saheli also have been isolated and sequenced. Our results presented here indicate that the sequence of the nolR gene is well conserved among all the (sino)rhizobia strains investigated. S. teranga was the only exception because it appears that nolR is not present in this bacterium. The fact that all attempts to amplify the complete S. saheli and S. xinjiangense nolR genes failed indicates that the S. fredii sequences used as primers (external to the nolR coding sequence) are poorly conserved, or even absent, in these two bacteria. In the presence of genistein, the introduction of multiple copies of nolR in S. fredii HH103 clearly produced a decrease of the expression of all the flavonoid-inducible genes tested in this work (Table 1). In addition, multiple copies of nolR also repressed the expression of nodD1, this gene being the only one that also is downregulated in the absence of flavonoids. The absence of NolR also had an effect on the expression of nod genes. In the absence of flavonoids, the β-galactosidase activity of the nolR mutant SVQ517 carrying plasmid pMP240 (it contains a transcriptional fusion between the R. leguminosarum bv. viciae nodA promoter and the lacZ gene) was nearly double that shown by the parental strain HH103 (pMP240). In the presence of genistein, such differences were not clearly observed. Studies carried out by Cren and associates (1995) showed that, in S. meliloti, NolR represses both nodD1 and nodD2 genes, causing a general decrease of nod gene expression. In addition, the S. meliloti NolR protein specifically binds to the promoter regions of nod operons involved in the synthesis of the Nod factor core, but not to those of nod operons related to the decoration of LCO because these operons lack NolR-binding boxes in their promoters. This differential regulation results in the preferential synthesis of fully decorated LCO. The situation appears to be different in S. fredii because putative NolR-binding sites are found in the promoter regions of two different operons, one comprising both common and specific nod genes (nodABCIJnolOnoeI) and the other one containing only genes involved in Nod factor decoration (nodZnoeLnolK). The repression caused in the expression of nod genes of S. fredii HH103 by multiple copies of nolR varies from 40 to 66%, if mutant SVQ123 is not included (Table 1). This mutant carries a Tn5-lacZ insertion into a putative ORF that is homologous to the y4pF gene of Rhizobium sp. NGR234 that encodes a putative transposase whose symbiotic relevance, if any, is unknown. Our results show that inactivation of the S. fredii nolR gene not only led to an increase of the amount of LCO detected but also to the synthesis of specific Nod factors that apparently are not produced by the parental strain HH103, such as those bearing N-methylations (Table 2). The possibility that the wild type strain HH103 also produces these new Nod factors, albeit at very low quantities, cannot be totally discarded. However, N-methylated Nod factors have not been identified among those produced by S. fredii HH103 when larger volumes (10 instead of 3 liters) were used (Gil-Serrano et al. 1997). Moreover, mutant SVQ517 carrying plasmid pMUS675 did not produce detectable amounts of N-methylated Nod factors. Hence, although S. fredii HH103-1 has the intrinsic capacity to produce N-methylated LCO, these Nod factors are produced only in the absence of a functional nolR gene. Krishnan and associates (1992) demonstrated that the inability of S. fredii USDA257 to nodulate Leucaena spp. was due to the lack of expression of the nodS gene, which is implied in the N-methylation of Nod factors in Rhizobium sp. NGR234 (Jabbouri et al. 1995). However, the production of N-methylated LCO by S. fredii strain SVQ517 did not habilitate this strain to nodulate Leucaena leucocephala (data not shown). The presence in S. fredii HH103 of a functional nodS gene, or another gene involved in N-methylation of LCO, has not been investigated. The fact that a wider range of Nod factors are detected in the HH103 nolR mutant (compared with its wild type, HH103) could be due to differences in the levels of repression of the different nodulation genes (Table 1). Variations in the relative amounts of the different LCO detected (Table 2) clearly are observed when multiple copies of nolR are present, which again suggest different levels of repression of the different nod genes involved in LCO synthesis. Inactivation of the noeL gene (which is involved in LCO fucosylation) also provokes the appearance of new LCO that were not detected in the wildtype strain (Lamrabet et al. 1999), although they were different from those observed in the nolR mutant. In addition, Tn5lacZ insertions upstream the noeI gene (which codes for a methyl-transferase) generated mutants in which the ratio of fucosylated/methyl-fucosylated Nod factors is increased 682 / Molecular Plant-Microbe Interactions (Madinabeitia et al. 2002), as is observed when HH103 carries multiple copies of nolR. The NolR protein belongs to a family of small bacterial regulatory proteins that comprises both positive and negative elements (Kiss et al. 1998). Proteomic analyses in S. meliloti have shown that nolR regulates proteins involved in different cellular processes such as the tricarboxilic acid cycle, heatand cold-shock responses, and protein synthesis (Chen et al. 2000). Although these results indicate that NolR acts as a global regulator, LCO production is the only symbiotic signal in which the role of NolR has been studied (Kiss et al. 1998; Kondorosi et al. 1991). In the present study, we provide evidence that S. fredii HH103 NolR not only regulates the production of LCO but also two other symbiotic signals: SR proteins and EPS. In S. fredii and Rhizobium sp. NGR234, SR proteins are involved in host range determination (Bellato et al. 1997; Viprey et al. 1998). Some S. fredii strains, such as USDA257, form nitrogen-fixing nodules with Asiatic soybean cultivars but fail to nodulate American cultivars (Buendía-Clavería and Ruiz-Sainz 1985; Keyser et al. 1982). This cultivar specificity somehow also is conditioned by SR proteins. In fact, some mutations into the nolXWBTUV locus (it encodes part of the TTSS and also some of the SR proteins) of S. fredii USDA257 abolish SR protein production and enable the bacteria to establish effective nodulation with American soybean cultivars (Krishnan et al. 1995; Meinhardt et al. 1993). Paradoxically, S. fredii strains, such as HH103, that naturally form nitrogen-fixing nodules with American soybean cultivars also secrete SR proteins (Lyra 2001). In the presence of flavonoids, transcriptional activation of Rhizobium sp. NGR234 nod genes involved in LCO production occurs earlier than that of the nolXWBTUV locus, suggesting a possible role of rhizobial SR proteins in advanced stages of root infection (Viprey et al. 1998). This hypothesis is supported by the fact that NolX, one of the S. fredii USDA257 SR proteins, has been localized in the infection threads of developing soybean and cowpea nodules (Krishnan 2002). In Rhizobium sp. NGR234, inactivation of nolX (also called nopX, for nodulation outer protein) has little effect on NopL and NopA secretion but greatly affected the interaction of this strain with many plant host tested (Marie et al. 2003). SR proteins from S. fredii and Rhizobium sp. NGR234 are produced in a flavonoid and nodD1 dependent manner. Genes involved in the TTSS of Rhizobium sp. NGR234 and B. japonicum lack nod boxes but are positively regulated by the product of ttsI (previously named y4xI) which is under the control of a nod box (Krause et al. 2002; Viprey et al. 1998). In S. fredii HH103, NolR represses different genes involved in SR protein production, such as nolX (encoding for an SR protein), and rhcQ and rhcJ (encoding for proteins of the TTSS) (Table 1). The promoter region of HH103 ttsI contains an NolR-binding site, which probably contributes to the dramatic reduction observed in the amount of SR proteins secreted by this strain in the presence of plasmid pMUS675. Only one S. fredii HH103 gene (nolX) coding for SR proteins has been identified (Bellato et al. 1997). Whether NolR affects the transcription of other genes coding for SR proteins remains to be investigated. Rhizobial surface polysaccharides are of crucial importance in the infection process of legume roots (Fraysse et al. 2003; Perret et al. 2000). Although EPSmutants of S. fredii are fully effective on G. max (Kim et al. 1989), some evidence indicates that it could be of biological significance: i) the presence of flavonoids reduces the amount of EPS produced by S. fredii USDA193 (Dunn et al. 1992); ii) the presence of multiple copies of nodD2 reduces EPS production by S. fredii USDA191 (Appelbaum et al. 1988; Machado and Krishnan 2003); and iii) USDA191 nodD1 or nodD2 mutants produced significantly more EPS than the wild-type strain, this effect being stronger in the nodD2 mutant. Results presented in this work add two new evidences that link EPS production with elements of the nod gene regulatory circuit. First, the amount of EPS recovered from S. fredii HH103 cultures is reduced in a NolR– background (mutant SVQ517), but increased if nolR is present in a multicopy vector (HH103-1 or SVQ517 carrying pMUS675); and second, the negative effect that genistein exerts on bacterial EPS production is attenuated by the presence of nolR. We also have shown that the hesB gene of S. fredii is repressed by NolR. This gene contains both nod and nif boxes in its promoter region, suggesting that hesB might act in late steps of the symbiotic interaction. Previous reports have described a negative effect of nolR mutations on the symbiotic properties of S. meliloti and R. leguminosarum bv. viciae with their leguminous host plants (Kiss et al. 1998; Kondorosi et al. 1989). In this article, we show that either inactivation or overexpression of nolR impairs the symbiotic interaction of S. fredii HH103 with soybean plants. Surprisingly, the symbiotic performance of mutant SVQ517 with V. unguiculata was similar to that shown by strain HH103 and the presence of plasmid pMUS675 in HH103-1 and SVQ517 produced a general improvement of the symbiotic parameters analyzed. In conclusion, our studies demonstrate the presence of nolR in all the S. fredii strains so far investigated and also in other Sinorhizobium spp. (such as S. saheli and S. xinjiangense). NolR in S. fredii HH103 clearly influences the production of at least three different symbiotic signals (Nod factors, SR proteins, and EPS) and also exerts a significant impact on the bacterial symbiotic capacity that varies with the host legume tested. Because the symbiotic signals affected by NolR act at different steps of the nodulation process, nolR appears to exert its influence at different stages of nodule formation. The regulating effect of NolR appears to be the result of the repression of genes that are directly involved in the synthesis of signals (such as nodA) or through the repression of other regulating genes that act as activators or repressors. The reduction in the production of LCO and the increase in EPS production would be carried out by repressing nodD1, and possibly nodD2, whereas the repression of the secretion of SR proteins in the presence of multiple copies of nolR could be achieved by repressing nodD1, and probably ttsI. This overall alteration of symbiotic signals, together with the relative importance of each signal alteration for the nodulation process with G. max and V. unguiculata, would account for the differences in symbiotic performance observed for the different bacterial strains tested. All these results indicate that it would be worthwhile to investigate whether nolR influences other symbiotic signals, such as KPS. MATERIALS AND METHODS Microbiological techniques. The bacterial strains and plasmids used in this work are described in Table 4. Sinorhizobium strains were grown at 28°C on tryptone yeast medium (Beringer 1974), yeast extract/mannitol (YM) medium (Vincent 1970), or B– medium (Spaink et al. 1992). Escherichia coli was cultured on LuriaBertani (LB) medium (Sambrook et al. 1989) at 37°C. When required, the media were supplemented with the appropriate antibiotics (µg ml-1) as described by Lamrabet and associates (1999). Genistein was dissolved in ethanol and used at 1 µg ml–1 (3.7 µM). Plasmids were transferred from E. coli to rhizobia by conjugation as described by Simon (1984). Vol. 17, No. 6, 2004 / 683 DNA manipulations. Recombinant DNA techniques were performed according to the general protocols of Sambrook and associates (1989). DNA sequencing was done by using the dideoxy-chain termination method (Sanger et al. 1977). PCR amplifications were performed according to protocols described by Saiki (1990). Primers used were: nolRint-f (5′ GAGGAAGCCGAAATAGCAG), nolRint-r (5′ GAGTCGGACGAGCTCGAATA), nolRupst (5′ TATGCTACCCCCAATTCTTGC), nolRdwst (5′ GAAAAAGC CCCGCGATTGCT), and nolR-r (5′ CCTGCGGCTTTTCGCT TCTCA). DNA and deduced protein sequences were analyzed with the UWGCG program (Devereux et al. 1984). The collection of HH103-1 derivatives carrying Tn5-lacZ insertions into flavonoid-inducible genes was generated as previously described (Lamrabet et al. 1999; Madinabeitia et al. 2002). Construction of strain SVQ502 (=HH103 Rifr nodD1:: lacZ∆p-Gmr) is described by Vinardell and associates (2004). The HH103 nolR derivative was generated as follows: plasmid pMUS672 was digested with NcoI, whose unique site is located in the middle of the nolR gene, filled with the Klenow enzyme, and fused to a 2-kb SmaI fragment carrying the Ω interposon (Spcr Strr) (Prentki and Krisch 1984). The plasmid generated, pMUS728, was digested with EcoRI and the 4.5-kb fragment carrying the nolR::Ω fusion was subcloned into the unique EcoRI site of pK18mob. The resulting plasmid, called pMUS735, was transferred to HH103-1 and Spcr Kms transconjugants were selected in order to identify putative double recombinants in which the wild-type nolR gene had been substituted by the mutated copy of this gene. Homogenization of the nolR::Ω in several candidates was confirmed by hybridization, and strain SVQ517 was selected for further studies. The 4.5-kb EcoRI fragment from pMUS728 carrying the nolR::Ω fusion also was subcloned into the broad-host-range vector pMP92, rendering plasmid pMUS810. Measurement of β-galactosidase activity. Assays for bacterial β-galactosidase activity in liquid YM medium were as described by Miller (1972). At least three independent experiments were performed, in duplicate, for each strain. Analysis of Nod factors. LCO from S. fredii strains grown in B– minimal medium were analyzed by RPTLC as described by Spaink and associates (1992). For chemical analyses, LCO were extracted and purified from 3 litters of bacterial cultures grown in B– minimal medium supplemented with genistein, as described by Lamrabet and associates (1999). Glycosyl composition and FABMS analyses were carried out as described by Madinabeitia and associates (2002). Analysis of extracellular proteins. Extracellular proteins from S. fredii strains were recovered from 50 ml of YM bacterial cultures grown on a orbital shaker (180 rpm) for 48 h (approximately 109 bacteria ml–1). Cultures Table 4. Bacterial strains and plasmids Strain or plasmid Derivation and relevant properties Source or reference Sinorhizobium fredii HH103-1 HH103 StrR Buendía-Clavería et al. (1989) SVQ269 HH103 RifR Madinabeitia et al. (2002) SVQ116 SVQ269 nodA::Tn5-lacZ Buendía-Clavería et al. (2003) SVQ118 HH103-1 nolX::Tn5-lacZ Bellato et al. (1997) SVQ120 HH103-1 rhcQ::Tn5-lacZ Madinabeitia et al. (2000) SVQ121 HH103-1 nolO::Tn5-lacZ Madinabeitia et al. (2002) SVQ123 HH103-1 carrying a Tn5-lacZ insertion into a locus homologous to y4pF of Rhizobium sp. NGR234 N. Madinabeitia SVQ124 HH103-1 hesB::Tn5-lacZ Espuny et al. (2000) SVQ287 HH103-1 noeL::Tn5-lacZ Lamrabet et al. (1999) SVQ288 HH103-1 rhcJ::Tn5-lacZ Lyra (2001) SVQ296 HH103-1 carrying a Tn5::lacZ insertion elsewhere M. T. Cubo SVQ502 SVQ269 nodD1::lacZ∆p-GmR Vinardell et al. (2004) SVQ517 HH103-1 nolR::Ω This work 042B(s)R 042B(s) RifR A. Buendía-Clavería USDA205 Wild-type strain Keyser et al. (1982) USDA257 Wild-type strain Keyser et al. (1982) SMH12 Wild-type strain Rodríguez-Navarro et al. (1996) S. saheli USDA4102 Wild-type strain de Lajudie et al. (1994) S. teranga USDA4101 Wild-type strain de Lajudie et al. (1994) S. xinjiangense CCBAU110 Wild-type strain Chen et al. (1988) Rhizobium sp. NGR234 Wild-type strain Trinick (1980) Escherichia coli DH5α supE44, ∆lacU169, hsdR17, recA1, endA1, gyrA96 , thi-1, relA1, NxR Stratagene Plasmids pBluescript II SK+ Cloning and sequencing vector, ApR Stratagene pHP45Ω ApR vector containing the Ω interposon (SpcRStrR) Prentki and Krisch (1984) pK18mob Cloning vector, KmR Schafer et al. (1994) pMP92 Broad-host-range cloning vector; IncP; TcR Spaink et al. (1987) pRK2013 Helper plasmid, KmR Figurski and Helinski (1979) pMUS671 Cosmid pLAFR1 carrying the nolR gene of S. fredii HH103 This work pMUS672 pBluescript carrying a 2,547-bp EcoRI fragment from pMUS671 containing HH103 nolR This work pMUS675 pMP92 carrying a 2,547-bp EcoRI fragment from pMUS671 containing HH103 nolR This work pMUS728 pMUS672 carrying the Ω interposon subcloned into the NcoI site of nolR. This work pMUS735 pK18mob carrying a 4.5-kb EcoRI fragment containing HH103 nolR::Ω derived from pMUS728 This work pMUS810 pMP92 carrying a 4.5-kb EcoRI fragment containing HH103 nolR::Ω derived from pMUS728 This work 684 / Molecular Plant-Microbe Interactions were centrifuged at 10,000 × g for 10 min and three volumes of acetone were added to the supernatants. The resultant mixtures were stored at –20°C for 24 h and centrifuged at 22,000 × g for 20 min. Dried pellets were resuspended in 500 µl of sample buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS [wt/vol], 10% glycerol [wt/vol], 5% β-mercaptoethanol [wt/vol], 0.001% bromophenol blue [wt/vol]). Extracellular proteins were separated by SDS-PAGE using the discontinuous buffer system of Laemmli (1970). Electrophoresis were performed in 0.75-mm-thick gels (15% acrylamide for the separation gel; 4% acrylamide for the stacking gel) in a Protean electrophoresis system (Bio-Rad, Hercules, CA, U.S.A.) at 25°C and 150 V. The amount of each sample loaded into the wells varied (25 to 50 µl) according to the bacterial density estimated for each culture. Gels were silver stained as described by Switzer and associates (1979). Studies of external polysaccharides. For extraction of EPS from liquid cultures, S. fredii strains were grown in 50 ml of liquid B– minimal medium on an orbital shaker (180 rpm) for 96 h (approximately 5 to 7 x 109 bacteria ml–1). When required, the media were supplemented with tetracycline (1 µg/ml), genistein (1 µg/ml), or both. After centrifugation at 10,000 × g for 10 min, supernatants were mixed with three volumes of acetone and the EPS was recovered with a glass-rod, dried at 60°C for 3 h, and weighed. This analysis was repeated twice, giving similar results. To investigate EPS production on solid B– medium, rhizobial strains were grown for 120 h at 28°C followed by 48 h at room temperature. LPS extraction, separation on SDS-PAGE, and silver staining were performed as previously described (Gil-Serrano et al. 1999). Plant assays. S. fredii strains were tested on G. max cv. Williams and V. unguiculata (cowpea) as described by Buendía-Clavería and associates (1989). Each Leonard jar contained two soybean plants or one cowpea plant. Each plant was inoculated with approximately 5 × 108 bacteria. Plants were grown for 52 days with a 16-h photoperiod at 25°C in the light and 18°C in the dark. Plant tops and nodules were dried at 70°C for 48 h and weighed. Bacteria were isolated from surface-sterilized nodules as previously reported (Buendía-Clavería et al. 1989). For competition studies on G. max cv. Williams, each plant was inoculated with approximately 3 × 108 bacteria of each coinoculant, HH103-1 (Strr) and SVQ517 (StrrSpcr). Nodule occupancy by strain SVQ517 was determined in 125 nodules by assessing the presence of the Ω interposon (presence of the Spcr marker). Experiments to determine the nodulation rate of soybean plants inoculated with strains HH103-1 or SVQ517 were carried out in mini-Leonard jars (approximately 200 ml for the upper part containing vermiculite and 170 ml for the reservoir containing the plant nutritive solution). At least six plants inoculated with each strain were analyzed for nodulation at each time point. ACKNOWLEDGMENTS This work was supported by CICYT grants BIO99-0614-C03 and BOS2002-04164-C03-02. J. M. Vinardell was supported by an IHP Return Fellowship from the Marie Curie program (EC). We are grateful to Rocío Gutiérrez for technical assistance. LITERATURE CITED Appelbaum, E. R., Thompson, D. V., Idler, K., and Chartrain, N. 1988. Bradyrhizobium japonicum USDA191 has two nodD genes that differ in primary structure and function. J. Bacteriol. 170:12-20. Bellato, C., Krishnan, H. B., Cubo, T., Temprano, F., and Pueppke, S. G. 1997. The soybean cultivar specificity gene nolX is present, expressed in a nodD-dependent manner, and of symbiotic significance in cultivarnonspecific strains of Rhizobium (Sinorhizobium) fredii. Microbiology 143:1381-1388. Beringer, J. E. 1974. R factor transfer in Rhizobium leguminosarum. J. Gen. Microbiol. 84:188-198. Buendía-Clavería, A. M., Chamber, M., and Ruiz-Sainz, J. E. 1989. A comparative study of the physiological characteristics, plasmid content and symbiotic properties of different Rhizobium fredii strains. Syst. Appl. Microbiol. 12:203-209. Buendía-Clavería, A. M., Moussaid, A., Ollero, F. J., Vinardell, J. M., Torres, A., Moreno, J., Gil-Serrano, A. M., Rodríguez-Carvajal, M. A., TejeroMateo, P., Peart, J. L., Brewin, N. J., and Ruiz-Sainz, J. E. 2003. A purL mutant of Sinorhizobium fredii HH103 is symbiotically defective and altered in its lipopolysaccharide. Microbiology 149:1807-1818. Buendía-Clavería, A. M. and Ruiz-Sainz, J. E. 1985. Isolation of mutants of fast-growing soybean strains that are effective on commercial soybean cultivars. Physiol. Plant. 64:507-512. Chen, H., Higgins, J., Kondorosi, E., Kondorosi, A., Djorjevic, M. A., Weinman, J. J., and Rolfe, B. G. 2000. Identification of nolR-regulated proteins in Sinorhizobium meliloti using proteome analysis. Electrophoresis 21:3823-3832. Chen, W. X., Yang, G. H., and Li, J. L. 1988. Numerical taxonomic study of fast-growing soybean rhizobia and a proposal that Rhizobium fredii be assigned to Sinorhizobium gen. nov. Int. J. Syst. Bacteriol. 38:392397. Cren, M., Kondorosi, A., and Kondorosi, E. 1994. An insertional point mutation inactivates NolR repressor in Rhizobium meliloti 1021. J. Bacteriol. 176:518-519. Cren, M., Kondorosi, A., and Kondorosi, E. 1995. NolR controls expression of the Rhizobium meliloti nodulation genes involved in the core Nod factor synthesis. Mol. Microbiol. 15:733-747. Cullimore, J. V., Ranjeva. R., and Bono, J. J. 2001. Perception of lipochitin oligosaccharidic Nod factors in legumes. Trends Plant. Sci. 6:2430. de Lajudie, P., Willems, A., Pot, B., Dewettinck, D., Maestrojuan, G., Neyra, M., Collins, M. D., Dreyfus, B., Kersters, K., and Gillis, M. 1994. Polyphasic taxonomy of Rhizobia: emendation of the genus Sinorhizobium and description of Sinorhizobium meliloti comb. nov., Sinorhizobium saheli sp. nov., and Sinorhizobium teranga sp. nov. Int. J. Syst. Bacteriol. 44:715-733. Devereux, J., Haeberli, P., and Smithies, O. 1984. A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res. 12:387395. Dunn, M. F., Pueppke, S. G., and Krishnan, H. B. 1992. The nod gene inducer genistein alters the composition and molecular mass distribution of extracellular polysaccharides produced by Rhizobium fredii USDA193. FEMS (Fed. Eur. Microbiol. Soc.) Microbiol. Lett. 97:107112. Economou, A., Hamilton, W. D. O., Johnston, A. W. B., and Downie, J. A. 1990. The Rhizobium nodulation gene nodO encodes a Ca2+-binding protein that is exported without N-terminal cleavage and is homologous to haemolysin and related proteins. EMBO (Eur. Mol. Biol. Organ.) J. 9:349-354. Espuny, M. R., Bellogín, R. A., Cubo, M. T., Lamrabet, Y., Ollero, F. J., Ruiz-Sainz, J. E., Temprano, F., and Vinardell, J. M. 2000. A hesB mutant of Sinorhizobium fredii HH103 shows reduced nodulation ability. Page 188 in: Fourth European Nitrogen Fixation Conference. Book Abstr. J. Olivares and A. J. Palomares, eds. Seville, Spain. Fellay, R., Hanin, M., Montorzi, G., Frey, J., Freiberg, C., Golinowski, W., Staehelin, C., Broughton, W. J., and Jabbouri, S. 1998. nodD2 of Rhizobium sp. NGR234 is involved in the repression of the nodABC operon. Mol. Microbiol. 27:1039-1050. Figurski, D. H., and Helinski D. R. 1979. Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc. Natl. Acad. Sci. U.S.A. 76:1648-1652. Fraysse, N., Couderc, F., and Poinsot, V. 2003. Surface polysaccharide involvement in establishing the rhizobium-legume symbiosis. Eur. J. Biochem. 270:1365-1380. Geurts, R., and Bisseling, T. 2002. Rhizobium Nod factor perception and signaling. Plant Cell (Suppl.) S239-S249. Gil-Serrano, A. M., Franco-Rodríguez, G., Tejero-Mateo, P., ThomasOates, J., Spaink, H. P., Ruiz-Sainz, J. E., Megías, M., and Lamrabet, J. 1997. Structural determination of the lipo-chitin oligosaccharide nodulation signals produced by Rhizobium fredii HH103. Carbohydr. Res. 303:435-443. Gil-Serrano, A. M., Rodríguez-Carvajal, M. A., Tejero-Mateo, P., Espartero J. L., Menéndez, M., Corzo, J., Ruiz-Sainz, J.E., and Buendía-Clavería, A. M. 1999. Structural determination of a 5-