MPMI Vol. 18, No. 11, 2005, pp. 1175–1185. DOI: 10.1094 / MPMI -18-1175. © 2005 The American Phytopathological Society A ClC Chloride Channel Homolog and Ornithine-Containing Membrane Lipids of Rhizobium tropici CIAT899 Are Involved in Symbiotic Efficiency and Acid Tolerance Keilor Rojas-Jiménez,1 Christian Sohlenkamp,2 Otto Geiger,2 Esperanza Martínez-Romero,2 Dietrich Werner,1 and Pablo Vinuesa1,2 1FB Biologie der Philipps-Universität, FG Zellbiologie und Angewandte Botanik, Karl-von-Frisch-Str., D-35032 Marburg, Germany; 2Centro de Ciencias Genómicas, Universidad Nacional Autónoma de México, Cuernavaca, Morelos, México Submitted 13 March 2005. Accepted 22 June 2005. Rhizobium tropici CIAT899 is highly tolerant to several two pH units below the surrounding bulk soil (Marschner environmental stresses and is a good competitor for nod- 1995). In addition, the presence of antibacterial molecules ule occupancy of common bean plants in acid soils. Ran- and the strong competence between microorganisms for nu- dom transposon mutagenesis was performed to identify trients constitute further stress factors that could constrain novel genes of this strain involved in symbiosis and stress nodulation (Jjemba 2001; Walker et al. 2003). Finally, rhizo- tolerance. Here, we present a genetic analysis of the locus bial bacteroids also face an acidic environment in the peri- disrupted by the Tn5 insertion in mutant 899-PV9, which bacteroid space, which has been estimated to be up to two lead to the discovery of sycA, a homolog of the ClC family pH units more acidic than the plant cell cytosol (Udvardi and of chloride channels and Cl–/H+ exchange transporters. A Day 1997). Bacteroids also face osmotic and oxidative nonpolar deletion in this gene caused serious deficiencies stresses as well as microaerobiosis in the symbiosomes (Day in nodule development, nodulation competitiveness, and et al. 2001; Nogales et al. 2002). Acid stress is, therefore, a N2 fixation on Phaseolus vulgaris plants, probably due to common limiting factor all the way from the soil to the sym- its reduced ability to invade plant cells and to form stable biosome, which suggests the existence of different mecha- symbiosomes, as judged by electron transmission micros- nisms of adaptation. copy. A second gene (olsC), found downstream of sycA, is It is well-known that rhizobial species exhibit different homologous to aspartyl/asparaginyl β-hydroxylases and levels of tolerance to acidity (Munns et al. 1979; Graham et modifies two species of ornithine-containing lipids in vivo, al. 1982, 1994). However, the genetic and physiological presumably by hydroxylation at a still-unknown position. bases of this acid tolerance are still not clear. Two mecha- A mutant carrying a nonpolar deletion in olsC is symbi- nisms related to the acid-tolerance response have been identi- otically defective, whereas overexpressed OlsC in the fied in rhizobia and enterobacteria (Foster 1999; Merrell and complemented strain provokes an acid-sensitive pheno- Camilli 2002; O’Hara and Glenn 1994;). The first involves type. This is the first report of a ClC homolog being the synthesis of outer membrane proteins and changes in the essential for the establishment of a fully developed N2- structure of lipopolysaccharides, exopolysaccharides, and fixing root nodule symbiosis and of a putative β-hydroxy- fatty acids to enhance cell surface stability and to prevent lase that modifies ornithine-containing membrane lipids proton permeability (Ballen et al. 1990; Chen et al. 1993a; of R. tropici CIAT899, which, in turn, are contributing to Reeve et al. 1997). The second mechanism is related to the symbiotic performance and acid tolerance. maintenance of intracellular pH homeostasis (Chen et al. 1993b). Proton influx in low-pH environments such as that faced by Escherichia coli during gastric tract infection is In order to accomplish a successful symbiotic interaction counteracted in the cytoplasm by decarboxylation of amino with legumes, rhizobia have to cope with different stress con- acids to consume protons and antiporter activity to remove ditions they encounter in soil, the rhizosphere, and the products (Foster 1999; Merrell and Camilli 2002). Export of symbiosome. Soil acidity limits symbiotic N2 fixation and positively charged substrates could cause hyperpolarization crop productivity in many soils of the tropics and subtropics of the inner membrane; however, this process is prevented by (Aarons and Graham 1991; Hungria and Vargas 2000; Zahran Cl–/H+ exchangers of the ClC family, which act as electrical 1999). It causes nitrogen deficiency in legumes as it inhibits shunts (Accardi and Miller 2004; Chen 2005; Iyer et al. rhizobial growth, root infection, and bacteroid activity 2002;). Additional but yet-unknown mechanisms for acid tol- (Glenn et al. 1999; Munns et al. 1981). In the rhizosphere, erance might operate in rhizobia. plants secrete H+ and organic acids that acidify the soil up to Rhizobium tropici CIAT899 is highly tolerant to many envi- ronmental stresses and particularly to acidity. It is able to grow Corresponding author: P. Vinuesa; Av. Universidad s/n, col. Chamilpa on media acidified down to pH 4.0, and it is a good competitor Apdo. 565A; Telephone: +52 777 313 1697; Fax +52 777 317 5581; E- for nodule occupancy of Phaseolus vulgaris (common bean) mail: vinuesa@ccg.unam.mx and other hosts under acidic conditions (Graham 1992; Nucleotide sequence data from this study (3,761 bp) are available from the Martínez-Romero et al. 1991). Thus, R. tropici CIAT899 repre- GenBank database under accession number AY954450. sents a good model to search for genes involved in acid toler- Vol. 18, No. 11, 2005 / 1175 ance and to determine their role in symbiosis (Vinuesa et al. orf3. orf1 and orf4 were truncated at their 5′ ends. The genetic 2003). The current knowledge of pH-regulated genes in rhizo- analysis presented herein targeted orf2 and orf3, which are the bia is still poor, despite their agricultural relevance. Proteome ORFs that could be affected by the Tn5 insertion in mutant analyses using two-dimensional gel electrophoretic analysis 899-PV9 (Fig. 1). reveal differential protein synthesis after pH shifts (Aarons and In frame +2, at position 785, a 1,368-bp ORF (orf2) begins Graham 1991; Peick et al. 1999). Mutagenesis with the trans- that is predicted to encode a 48.6-kDa product that, accord- poson Tn5 and selection of mutants on acidified media was ing to homology searches with the BLASTP program, used to characterize acid-sensitive mutants in R. leguminosa- showed highly significant sequence similarity to ClC chlo- rum (Chen et al. 1993b). It has been shown that the R. tropici ride channel proteins from diverse β-proteobacteria gshB-like gene and Sinorhizobium meliloti actA, actP, exoH, Burkholderia cepacia (E value e–104, 49% identity, 64% simi- exoR, actS, actR, and phrR are essential for growth at low pH larity), Ralstonia solanacearum (E value e–103, 49% identity, (Glenn et al. 1999; Riccillo et al. 2000). More recently, 64% similarity), and the γ-proteobacteria Pseudomonas sy- Vinuesa and associates (2003) reported the isolation of addi- ringae (E value e–103, 45% identity, 62% similarity), Salmo- tional Tn5-induced acid-sensitive mutants of R. tropici nella typhimurium (E value 2e–55, 27% identity, 42% similar- CIAT899, all of which displayed symbiotically defective phe- ity), and E. coli (E value 4e–45, 20% identity, 35% similarity). notypes in terms of nodulation competitiveness, nodule devel- The product encoded by orf2 belongs to the voltage-gated opment, and N2 fixation on P. vulgaris. Strain 899-PV4 was chloride channel and antiporter protein family (COG0038). shown to carry a single Tn5 insertion at the 5′ end of atvA, an They are integral membrane proteins with inner membrane ortholog of the chromosomal virulence gene acvB of Agrobac- localization in gram-negative bacteria. The rhizobial ClC terium tumefaciens, which is required by the bacteria for both chloride channel homolog is predicted to have 10 transmem- a proficient interaction with plants and for acid tolerance brane helices, with both N- and C-terminal domains residing (Vinuesa et al. 2003). in the cytoplasm and containing the sequence motifs corre- In this study, we present a genetic analysis as well as a phe- sponding to the ion-binding site and gating region of the notypic characterization of the locus disrupted by the Tn5 in- solved E. coli and Salmonella typhimurium ClC protein struc- sertion in strain 899-PV9, which bears significant sequence tures (Fig. 1B, part I). This superfamily of ion channels and homology with the ubiquitous superfamily of ClC-Cl– channel antiporters is found in both prokaryotic and eukaryotic cells, proteins and Cl–/H+ antiporters (Accardi and Miller 2004; including most of the α-proteobacterial genomes sequenced Chen 2005), and of a gene located downstream that is involved to date. One of the two ClC paralogs found in E. coli, named in membrane lipid modifications, with relevance for acid toler- ClC-ec1, has been proposed to act as an electrical shunt for ance and symbiosis. an outwardly directed proton pump that is linked to amino acid decarboxylation as part of the extreme acid-resistance RESULTS response of this bacterium (Iyer et al. 2002). Protein sequence alignments and secondary structure analyses showed a high Isolation of cosmids that restore symbiotic proficiency conservation (Fig. 1B, parts I through III), including the R. in strain 899-PV9. tropici residue E129, which, in the homologous ClC chloride A symbiotically-defective mutant of R. tropici CIAT899, ob- channels from E. coli and S. typhymurium, is the one respon- tained by random transposon mutagenesis, was designated sible for gating, according to X-ray crystallographic studies 899-PV9. When mutant 899-PV9 was inoculated on P. vul- (Dutzler 2004; Dutzler et al. 2002). Therefore, based on the garis plants, nodules were not fully developed, were irregular highly significant sequence similarity of orf2 to ClC chloride in size, lacked lenticels, and did not express nitrogenase activ- channels and antiporters, we named the former locus sycA, ity. Southern blot analysis (not shown) confirmed that the ob- for symbiosis-assisting ClC homolog. The intergenic spacer served phenotype of this strain is due to a single Tn5 insertion, (IGS) between orf1 and orf2 was analyzed using the neural which was localized between nucleotides C1763 and T1764 of network for promoter prediction (NNPP) server, which located the sequence (AY954450) reported herein (Fig. 1A). Mapping a putative promoter sequence upstream of the sycA start codon of the insertion junction site was achieved by sequencing the (sites 727 to 772, r = 0.81). flanking regions of plasmid p899PV9ESK, which contains the A 845-bp-long ORF (orf3) starts in frame +1 at position 2,611 Tn5 insertion of strain 899-PV9 cloned as an EcoRI fragment, and is predicted to encode a 31.78-kDa product, with highly sig- using primer Tn5-77/58EB (Vinuesa et al. 2003), which reads nificant sequence similarity to a putative aspartyl/ outwardly from the Tn5 insertion sequence elements. asparaginyl β-hydroxylase from a Mesorhizobium sp. (E value Several symbiotically proficient transconjugants were recov- 9e–90, 64% identity, 76% similarity) and Azotobacter vinelandii ered after complementation of the mutant strain with a cosmid (E value 4e–63, 49% identity, 65% similarity) and a lower simi- library. The cosmids isolated from these transconjugants were larity to lipid A-myristate β-hydroxylase (LpxO) from Salmo- identical in their restriction patterns, after digestion with dif- nella typhimurium (E value 5e–23, 35% identity, 51% similarity). ferent enzymes. One of these cosmids, named pPV9cos2, was LpxO has been shown to be responsible for the 2-hydroxylation digested with EcoRI, was transferred to a nylon membrane, of myristate in lipid A (Gibbons et al. 2000). Sequence analyses and was hybridized with digoxigenin (DIG)-labeled p899PV9- predicted a cytoplasmic localization for the Orf3, with N- and PCR. The resulting hybridizing fragment (approximately 6.2 C-terminal hydrophobic domains. The catalytic domain of this kb) was subcloned into pSK, yielding pPV9EG-SK, and was aspartyl/asparaginyl β-hydroxylase protein family is well con- subjected to DNA sequencing. served (Fig. 1C). In particular, histidine residue H164 is thought to be an iron ligand and, therefore, essential for the function of Sequence analyses and characterization of orf2 and orf3. the protein (Jia et al. 1994). A maximum likelihood phylogenetic Using different computer programs, an analysis of a 3,761- reconstruction based on selected protein sequences of this bp EcoRI-ClaI sequence complementing mutant PV9-899 (ac- family found in divergent bacteria (data not shown) revealed that cession number AY954450) revealed four open reading frames the putative aspartyl/asparaginyl β-hydroxylase from R. tropici (ORF) with high coding probability, as predicted by FrameD is located in a different clade than the LpxO protein from Salmo- (Fig. 1A). ORF 1 through 3 are transcribed in the same orien- nella typhimurium. This inference suggests that LpxO and Orf3 tation, while orf4 is transcribed convergently with respect to might have slightly different functions or substrate specificities, 1176 / Molecular Plant-Microbe Interactions although they clearly belong to the same protein superfamily RT-PCR of the IGS between sycA and olsC confirms (COG3555). The IGS between sycA and orf3 was analyzed that both genes are independently transcribed. using the NNPP server, which predicted two possible promoter An RT-PCR experiment was designed to determine the tran- sequences upstream of orf3 (sites 2,306 to 2,351, r = 0.98; 2,419 scriptional organization of sycA and olsC (Fig. 1D). For this to 2,464, r = 0.94). This promoter prediction is consistent with a purpose, RNA was isolated from R. tropici CIAT899 cells monocistronic organization of the transcript encoded by sycA, grown in PY broth (Noel et al. 1984) to the early stationary which is also supported by reverse transcription-polymerase phase and was used to synthesize cDNA with random hexam- chain reaction (RT-PCR) experiments, complementation analy- ers as primers. This cDNA was used as template for PCR am- ses, and the phenotype observed by two-dimensional thin-layer plification experiments with primers PV9-2000f and PV9- chromatography (2D-TLC) analyses of radiolabeled lipid ex- 2985r, which bind up- and downstream of the IGS region (Fig. tracts (discussed below). Based on the evidence gained from the 1A). No amplification product was detected (Fig. 1D), which latter experiments, we named orf3 olsC, for ornithine lipid syn- suggests that both ORF are independently transcribed under thesis gene C. these conditions and making it unlikely that the Tn5 insertion Fig. 1. A, Genetic and physical maps of the 3,761-bp EcoRI-ClaI region from Rhizobium tropici CIAT899 analyzed in this study (accession number AY954450). A, Selected restriction sites. Four open reading frames (ORF) (represented by arrows) were detected. The site of the Tn5 insertion located in sycA between nucleotides C1763 and T1764 is indicated by an open triangle. Nonpolar deletion mutants lacking the regions shown in white were generated in sycA and olsC. Predicted promoters are shown as thin arrows. The dotted line represents the intergenic spacer between sycA and olsC subjected to reverse transcriptase-polymerase chain reaction (RT-PCR) analysis (shown in D). The dashed line shows the location of the 1.66-kb BamHI-EcoRI fragment cloned into pBBR-MCS5 and used to complement strain 899-olsC∆1. B, Transmembrane topology predicted by Predictprotein for SycA (panel I); the conserved residue E129 (marked with asterisk) is located in the gating region of EriC (Dutzler et al. 2003). Panels II and III show hydrophobicity plots generated by TmPred for the R. tropici CIAT899 ClC homolog sycA (II) compared with that for the ClC exchange transporter (EriC) from Salmonella typhimurium (III). C, Partial sequence alignment of R. tropici, Mesorhizobium sp. (ZP 00193099), Brucella melitensis (NP 539381), Azotobacter vinelandii (ZP 00090437) putative aspartyl/asparaginyl β-hydroxylase sequences, and LpxO (Fe2+/α-ketoglutarate-dependent dioxygenase) from Salmonella typhimurium (AAF87784). Residue His164 is highly conserved as part of the catalytic domain (Jia et al. 1994). Identical residues are underlined in black, and similar resi- dues are shaded in gray. D, The RT-PCR experiment shows no PCR amplification of the intergenic spacer (IGS) between sycA and olsC (lane 1), which sug- gests that both ORF are independently transcribed under these conditions. No amplification in the negative control (lane 2). Positive controls include the PCR amplification of an internal fragment from olsC (lane 3) and the 16S rDNA gene (lane 4), when using cDNA as template, or amplification of the IGS region, when using genomic DNA as template (lane 5). Vol. 18, No. 11, 2005 / 1177 in sycA has a polar effect on olsC. However, the IGS region Construction of nonpolar deletions in sycA and olsC could be amplified with PV9-2000f and PV9-2985r when ge- and phenotypic characterization of the mutant strains. nomic DNA was used as template. The presence of an internal We generated a nonpolar mutation in sycA by deletion of a fragment of olsC and 16S rDNA in the cDNA template was 991-bp fragment that embraces nearly the whole gene, includ- confirmed by PCR amplification of these genes, demonstrating ing the predicted gating region. To construct this mutant, plas- a proper cDNA synthesis. In the negative control without re- mid pKR∆02 was transferred into CIAT899. Double recombi- verse transcriptase, no amplification product was observed, nants were selected for the loss of sensitivity to sucrose, and which excludes the possibility of contamination by R. tropici the resulting deletion was confirmed by PCR with primers genomic DNA. PV9-524f-H and PV9-2272r-E (data not shown). The mutant obtained was designated 899-sycA∆1. This strain was able to grow in 20E or PY media acidified to pH 4.5 at similar rates as the parent strain (Fig. 2A). However, the nodules it formed on bean plants were poorly developed (21 days postinoculation [dpi]), lacked lenticels, were whitish (Fig. 3A and B) and pre- sented a 14-fold decrease in nitrogen fixation in comparison with the parent strain, as revealed by the acetylene reduction assay (Fig. 2B). The nodulation competitiveness of 899- sycA∆1 against the gusA-tagged reporter strain CIAT899-G1 in a 10:1 coinoculation experiment (in favor of the mutant) showed that only 5.9% of the nodules were occupied by the former (Fig. 2C). Coinoculations of CIAT899 and CIAT899- G1 resulted in percentages of GUS-expressing nodules that re- flected the proportion of gusA-tagged reporter cells in the in- oculum mixture, as previously reported (Vinuesa et al., 2003). The symbiotic phenotypes displayed by this strain correspond to those observed in mutant 899-PV9. Further analyses using light and transmission electron microscopy (TEM) revealed that both mutants were able to enter the nodules but were un- able to form stable symbiosomes. TEM micrographs of nod- ules induced by 899-sycA∆1 showed poor invasion of plant cells, accumulation of polyhydroxybutyrate granules within Fig. 2. Phenotypes displayed by several Rhizobium tropici strains used in this study. A, growth of R. tropici strains on PY media at pH 4.5. Values are the mean ± SD of four independent experiments. B, Mean acetylene reduction of nodulated roots inoculated with strains 899-PV9, 899- Fig. 3. Macroscopic aspect of nodules induced by strains A, CIAT899, B, sycA∆1, and 899-olsC∆1 in comparison to the parental strain CIAT899. mutant 899-sycA∆1, and C, mutant 899-olsC∆1 on Phaseolus vulgaris Values are the mean ± SD of three repetitions in two independent experi- plants (21 days postinoculation). Electron micrographs reveal drastic dif- ments. C, Percentage of nodules occupied by mutant strains (black) ferences in the levels of invasion presented by D, parent strain and E, mu- against reporter strain CIAT899-G1 (gray) in a 10:1 coinoculation experi- tant 899-sycA∆1. Notice the accumulation of polyhydroxybutyrate gran- ment (in excess of mutants), using low-inoculum titers (approximately 500 ules in bacteroids of the latter. F, Mutant 899-olsC∆1 also shows reduced CFU per plantlet). levels of invasion, as revealed by light microscopy. 1178 / Molecular Plant-Microbe Interactions bacteroids, and presence of amyloplasts in the host cells, carry the deletion, by PCR with primers PV9-2571f-E and whereas the parent strain was able to fully invade plant cells PV9-3622r-H (data not shown). Mutant 899-olsC∆1 was able (Fig. 3D and E). The symbiotic proficiency of mutant 899- to grow in 20E or PY media acidified to pH 4.5 at a similar sycA∆1 was restored when cosmid pPV9cos2 was provided in rate as the parent strain (Fig. 2A). Nodules of bean plants in- trans (data not shown). oculated with this strain (21 dpi) were poorly developed (Ndv–) A partial deletion (211 bp long) of the region containing the and lacked lenticels (Fig. 3C). These nodules also showed re- predicted catalytic domain of OlsC was generated using plas- duced levels of nitrogen fixation (about 50%), when compared mid pKR∆03, following the same procedure mentioned for with the wild-type strain CIAT899, as determined by acetylene sycA. The resulting strain (899-olsC∆1) was confirmed to reduction assays (Fig. 2B). When 899-olsC∆1 was coinocu- Table 1. Membrane lipid composition of Rhizobium tropici CIAT899 wild type, mutant 899-olsC∆1, complemented strain 899-olsC∆1/pBBR-1,6BE, and mutant containing the empty vector 899-olsC∆1/pBBR-MCS5a Composition (%total 14C) Lipid CIAT899 899-olsC∆1 899-olsC∆1/pBBR-1,6BE 899-olsC∆1/899-olsC∆1/ pBBR-MCS5PC 40.3 ± 4.7 47.8 ± 6.0 32.5 ± 1.5 37.4 ± 5.5 Phosphatidylglycerol (PG) 11.9 ± 2.1 10.7 ± 1.9 12.7 ± 1.3 11.2 ± 0.2 Cardiolipin (CL) 2.9 ± 2.0 2.6 ± 0.8 2.9 ± 0.3 3.2 ± 1.4 Phosphatidylethanolamine (PE) 23.4 ± 3.8 16.3 ± 4.9 33.4 ± 0.4 26.2 ± 1.7 S1b 2.7 ± 2.7 7.2 ± 1.8 n.d. 10.7 ± 3.0 S2 4.5 ± 1.9 15.3 ± 1.1 0.3 ± 0.2 11.3 ± 0.8 P1 6.3 ± 3.0 n.d. 6.8 ± 2.9 n.d. P2 8.1 ± 4.9 n.d. 11.4 ± 1.4 n.d. a Values are the mean ± standard deviations of three independent experiments. n.d. = not detected. b S1, S2, P1, and P2 denote four additional components detected in the chromatogram of R. tropici CIAT899. Fig. 4. Membrane lipid analysis of Rhizobium tropici strains. Separation of [14C]acetate-labeled lipids from A, R. tropici CIAT899, B, mutant 899-olsC∆1, and C, complemented mutant 899-olsC∆1/pBBR-1,6BE as well as of D, [14C]ornithine-labeled lipids from R. tropici CIAT899, using two-dimensional thin- layer chromatography. The lipids cardiolipin (CL), phosphatidylglycerol (PG), sulphoquinovosyl diacylglycerol (SL), phosphatidylethanolamine (PE), dimethylphosphatidylethanolamine (DMPE), and phosphatidylcholine (PC) are indicated. In A, B, and C, ovals surround lipid species S1 and S2, which presumably are substrates for the putative R. tropici β-hydroxylase to form the lipid products (P1 and P2) that are enclosed in rectangular boxes. Asterisks indicate ninhydrin-positive lipids. Vol. 18, No. 11, 2005 / 1179 lated with the gusA-tagged reporter strain CIAT899-G1 in a a significantly increased mean generation time (g = 9.2 h) in 10:1 ratio, only 25% of the nodules were occupied by the for- comparison to CIAT899 or the mutant 899-olsC∆1 with or mer, which displayed the same Ndv– phenotype as the nodules without the empty vector (g ≈ 2.9 h). These results suggest that induced by 899-olsC∆1 (Fig. 2C). Light micrographs of nod- the expression or overexpression (due to copy number) of the ules inoculated with mutant 899-olsC∆1 (Fig. 3F) revealed gene contained in this 1.66-kb fragment is responsible for the lower invasion levels than those achieved by the parent strain reduced growth under acidic conditions displayed by the com- CIAT899, although not as drastic as observed for mutant 899- plemented strain when carrying plasmid pBBR-1,6BE in trans. sycA∆1 (data not shown). Some membrane lipids Complementation of mutant 899-olsC∆1 of R. tropici CIAT899 are absent in mutant 899-olsC∆1. with pBBR-1,6BE. Lipid extracts from R. tropici CIAT899, mutant 899-olsC∆1 In order to complement mutant 899-olsC∆1, plasmid pBBR- carrying the 211-bp deletion in olsC, and the complemented 1,6BE (Fig. 1A) was constructed and provided in trans. A mutant 899-olsC∆1/pBBR-1,6BE were separated by 2D-TLC, 1,660-bp BamHI-EcoRI fragment containing olsC and the up- and individual lipids were quantified (Table 1). Rhizobial stream region with the predicted promoter region was PCR- membrane phospholipids like phosphatidylethanolamine (PE), amplified and cloned into pBBR-MCS5. To ensure that the dimethylphosphatidylethanolamine, cardiolipin, phosphatidyl- gene is expressed from its native promoter, the fragment was glycerol, sulphoquinovosyl diacylglycerol, and phosphatidyl- cloned in opposite transcriptional direction to the vector’s lacZ choline (PC) were identified based on their relative mobilities promoter. Plasmid pBBR-1,6BE was transferred into mutant and in comparison to the well-characterized lipid profile of S. 899-olsC∆1, and strain 899-olsC∆1/pBBR-1,6BE was tested meliloti 1021 (Gao et al. 2004; Weissenmayer et al. 2002). As on bean plants for symbiotic performance. The symbiotic pro- found for S. meliloti 1021, PC constitutes the major membrane ficiency and nitrogen fixation capacity of strain 899- lipid of R. tropici CIAT899. In addition to the above-men- olsC∆1/pBBR-1,6BE was restored to similar levels as exhib- tioned lipids, four additional components can be detected in ited by CIAT899. On near-neutral media (pH 6.8), the comple- the chromatogram of R. tropici CIAT899, labeled as S1, S2, mented strain 899-olsC∆1/pBBR-1,6BE (generation time (g) = P1, and P2 (Fig. 4A). Staining of developed 2D-TLC chroma- 2.8 h) grew similarly to CIAT899 (g = 2.4 h) or mutant 899- tograms with ninhydrin revealed that PE, S1, and P1 possess olsC∆1 with (g = 2.8 h) or without (g = 2.8 h) the empty vector primary amino groups (data not shown). The compound S1 (data not shown). In contrast, when the complemented strain shows the same relative mobility as ornithine-containing lipids was grown on acidified media at pH 4.5 (Fig. 2A), it presented (OL), which have been characterized previously in S. meliloti Table 2. Bacterial strains and plasmids used in this study Strain or plasmid Relevant characteristicsa Source or reference Rhizobium tropici strains CIAT899 Acid tolerant, Apr, Nalr Martínez-Romero et al. 1991 899-PV9 CIAT899 derivative (sycA::Tn5), symbiotically defective, Smr, Kmr Vinuesa et al. 2003 CIAT899-G1 gusA-tagged CIAT899 derivative carrying a single mTn5gusA30 insertion, used as Vinuesa et al. 2003 reporter strain in competition experiments, Smr , Spr 899-sycA∆1 CIAT899 carrying a 991 bp non-polar deletion in sycA This study 899-sycA∆1/pPV9cos2 899-sycA∆1 carrying the complementing cosmid of 899-PV9, Tcr This study 899-olsC∆1 CIAT899 carrying a 211-bp nonpolar deletion in olsC This study 899-olsC∆1/pBBR-1,6BE 899-olsC∆1 complemented with pBBR-1,6BE This study 899-olsC∆1/pBBR-MCS5 899-olsC∆1 carrying the vector pBBR-MCS5 This study Escherichia coli strains DH5α recA1, ∆lacU169, Φ80dlacZ∆M1 Stratagene, La Jolla, CA, U.S.A. S17-1 thi pro hsdR– hsdM+ recA, RP4 integrated in the chromosome, 2-Tc::Mu- Simon et al. 1983 Km::Tn7(Tpr/Smr) Plasmids pRK2013 Helper plasmid; Kmr Ditta et al. 1980 pK18mob Conjugative suicide vector with Kmr lacZ markers used for positive selection of Schäfer et al. 1994 single recombinants pK18mobsacB Conjugative suicide vector with Kmr lacZ and sacB markers used for positive Schäfer et al. 1994 selection of double recombinants pBBR-MCS5 Mobilizable broad host range cloning vector, Gmr Kovach et al. 1995 pCR II PCR cloning vector Invitrogen, Carlsbad, CA, U.S.A. pSUP1011 Mobilizable suicide plasmid for Tn5 mutagenesis Simon et al. 1985 pBluescript II SK (pSK) Standard cloning and sequencing vector, lacZ Apr Stratagene p899PV9ESK 12-kb EcoRI fragment from strain 899-PV9, containing the Tn5 insertion, cloned This study in pSK. Apr, Kmr p899PV9E-PCR pSK plasmid containing the flanking region of p899PV9ESK and used for DIG- This study labeling. Apr P859PV9E6-SK pSK construct containing the 6.2-kb EcoRV fragment hit by the Tn5 insertion in This study 899-PV9 pPV9cos2 Cosmid complementing 899-PV9, Tcr This study pKR∆02 Integrative mutagenizing plasmid based on pK18mobsacB used to construct strain This study 899-sycA∆1 pKR∆03 Integrative mutagenizing plasmid based on pK18mobsacB used to construct strain This study 899-olsC∆1 pBBR-1,6BE 1.66-kb BamHI-EcoRI fragment cloned into pBBR-MCS5 used for complementing This study mutation on strain 899-olsC∆1 a Ap, Nal, Sm, Km. Sp, Tc, and Tp = ampicillin, nalidixic acid, streptomycin, kanamycin, spectinomycin, tetracycline, and trimethoprim. 1180 / Molecular Plant-Microbe Interactions (Gao et al. 2004; Weissenmayer et al. 2002). Since OL is also mutated in this study is required for a proficient symbiotic a ninhydrin-positive compound, we suggest that S1 might be interaction with bean plants but, apparently, not for acid toler- the corresponding OL from R. tropici CIAT899. The lipid ance or growth under free-living conditions. However, at this composition of the Tn5-generated mutant 899-PV9 and that of point, we cannot define if the mutations made in sycA have a the deletion mutant 899-sycA∆1, both defective in the predicted direct or indirect effect on the symbiotic phenotype observed chloride channel protein, were indistinguishable from that of in strains 899-PV9 and 899-sycA∆1. At this point and based on the wild type (data not shown). Although the lipid composition the evidence gained from TEM data, we cannot state whether of mutant 899-olsC∆1 resembles that of the wild type (Fig. the mutation is affecting bacteroid release from infection 4B), compounds P1 and P2, which together comprised nearly threads, symbiosome proliferation, or stability. Mutations in 15% of the wild-type membrane lipids, are absent in mutant the second paralogous gene, the construction of a double mu- 899-olsC∆1 (Table 1). In contrast, the wild type has only minor tant, and analysis of transcriptional reporter gene fusions levels of S1 and S2 (7.2% of total membrane lipids), whereas would be very valuable to gain a better understanding of the in mutant 899-olsC∆1, these two lipids comprise 22.4% of the functions of these genes. total lipid detected. If the mutant is complemented in trans It has recently been shown that R. leguminosarum mutants with pBBR-1,6BE (Fig. 4C), again, P1 and P2 are formed in blocked in amino acid transporters, present poorly developed relatively high amounts (more than 18% of total membrane nodules and reduced N2 fixation and the bacteroids are satu- lipids), whereas S1 and S2 are practically absent from this rated with dicarboxylic acids and polyhydroxybutyrate gran- strain. These data suggest that OlsC is required for the forma- ules (Lodwig et al. 2003). Since the peribacteroid space is tion of lipids P1 and P2. acidic and the mutant 899-sycA∆1 displayed a similar pheno- Incorporation experiments with radiolabeled ornithine dem- type to that observed for the R. leguminosarum mutant, it is onstrated that from all the membrane lipids, in S. meliloti, only tempting to speculate that SycA might be involved in the adap- OL become labeled, indicating that ornithine is specifically tation of R. tropici bacteroids to the symbiosome’s milieu, and exclusively incorporated into OL (Gao et al. 2004). Analy- probably in relation with the electrophysiology of bacteroid sis of the lipid profile from R. tropici CIAT899 that had been membranes, which in turn may affect key aspects of cellular labeled with [1-14C]ornithine indicates that four compounds homeostasis like the internal pH of bacteroids or the transport have incorporated radiolabeled ornithine (Fig. 4D). The relative of metabolites across their cellular membranes. Since ClC mobilities of the radiolabeled compounds coincide with the channels and antiporters are highly selective for chloride ani- relative mobilities found for lipids S1, S2, P1, and P2. There- ons (Accardi et al. 2004; Chen 2005), this would imply that fore, we conclude that all four of these lipids are distinct classes Cl– could play a key role in symbiosome physiology. If so, it of OL of R. tropici CIAT899. remains to be uncovered. Only two previous works describe the phenotypes of micro- DISCUSSION bial cells carrying mutations in ClC homologs in relation to their interaction with eukaryotic hosts. Mutations in clc-a from In this work, we report two novel genes of Rhizobium trop- the human pathogenic yeast Cryptococcus neoformans resulted ici CIAT899 that were shown to be required for the establish- in attenuated virulence in a mouse cryptococcosis model. This ment of a fully developed N2-fixing symbiosis with bean attenuation resulted from the lack of expression of two impor- plants. The microscopy analyses performed on mutant 899- tant virulence factors, capsule and laccase (Zhu and Williamson sycA∆1, which carries a deletion in the putative chloride chan- 2003). In contrast, deletion of the single ClC ortholog found in nel encoded by sycA, suggests that the observed decrease in Vibrio cholerae enhanced intestinal colonization in infant nodule development and nitrogen fixation is probably due to mice. This gene was found to confer mild resistance to acid its failure to invade plant cells and to form stable symbio- when pH was adjusted with HCl but not with other acids (Ding somes. This is the first report showing that a rhizobial ho- and Waldor 2003). Clearly, much research is still needed to molog of the ClC family of Cl– channels and Cl–/H+ antiporters provide a basic understanding of the molecular and physiologi- is essential for the establishment of a fully proficient symbiotic cal functions of the diverse ClC homologs found across pro- interaction with its legume host but not for growth under free- karyotic phyla. living conditions. With the notable exception of E. coli, the It has been speculated that certain membrane lipids might be molecular and physiological functions of dozens of prokary- important for the establishment of a successful symbiosis (de otic ClC homologs recently uncovered by genome sequencing Rudder et al. 1997). This idea has been supported by the ob- projects are still unknown. It is worth noting that the ClC-ec1 servations that mutants of S. meliloti lacking phosphatidylcho- (or EriC) protein of E. coli was recently shown not to be an ion line (PC) are unable to form nitrogen-fixing nodules on alfalfa channel, but rather a H+/Cl– exchange transporter, demonstrat- (López-Lara et al. 2003; Sohlenkamp et al. 2003) and that mu- ing that the structural boundary separating transporters and tants of Bradyrhizobium japonicum with diminished levels of channels is not as clearcut as previously thought (Accardi and PC present a reduced number of bacteroids within infected Miller 2004; Chen 2005). The E. coli genome has two ClC ho- plant cells (Minder et al. 2001). Since PC comprises 50 to 60% mologs. When either one of these genes was individually de- of the lipids in membranes of S. meliloti or Bradyrhizobium leted, no notable phenotype was observed. However, a double japonicum and OL about 20% in R. tropici, one might expect knock-out strain displayed a dramatic reduction in cell survival that major changes in membrane lipid composition, either by and amino acids transport under acid shock (Iyer et al. 2002). mutations or by increased copy number of genes involved in We have recently cloned a second ClC-like paralog from their biosynthesis, could affect the structural properties of cell CIAT899 (as found in the genomes of many other α-Proteo- membranes and, as a consequence, the proper functioning of bacteria, including Agrobacterium tumefaciens C58, Brucella membrane-associated proteins. suis 1330, Brucella melitensis 16M, Bradyrhizobium japoni- R. tropici CIAT899 produces four different classes of OL cum USDA110, and Mesorhizobium loti MAFF303099 but not (S1, S2, P1, and P2). As lipid S1 is ninhydrin-positive and mi- S. meliloti 1021), which suggests that the ClC paralogs might grates in an identical way as the well-characterized OL (Geiger perform different functions in the cell under different physio- et al. 1999) from S. meliloti in 2D-TLC, we expect that S1 pre- logical conditions and might be differentially expressed. Our sents a similar or identical structure as sinorhizobial OL. In complementation results demonstrate that the paralog (sycA) contrast, the S2, P1, and P2 classes of OL, in this order, migrate Vol. 18, No. 11, 2005 / 1181 more slowly in both dimensions of 2D-TLC systems and, there- 1992), or in 20E medium (Werner et al. 1975) at 28°C. fore, must be increasingly more polar due to still unknown Acidic media at pH 4.5 were buffered with 25 mM Homopipes modifications by functional groups. (Research Organics, Cleveland, OH, U.S.A.). E. coli strains Mutant 899-olsC∆1 forms more S1 and S2 than wild type were grown in Luria-Bertani medium at 37°C (Sambrook et and lacks P1 and P2, whereas the complemented mutant 899- al. 1989). When needed, antibiotics were added at the fol- olsC∆1/pBBR-1,6BE presented mainly P1 and P2 and a near- lowing concentrations: kanamycin, 100 µg/ml; streptomycin, complete lack of S1 and S2. These data show that OlsC is re- 150 µg/ml; ampicillin (Ap), 50 µg/ml; tetracycline, 7.5 quired for the formation of P1 and P2 and suggest that OlsC µg/ml; gentamicin (Gm), 10 µg/ml; nalidixic acid (Nal), 20 might consume S1 and S2. This is consistent with a model in µg/ml. which the predicted β-hydroxylase encoded by olsC converts the ninhydrin-positive compound S1 to the ninhydrin-positive Random transposon mutagenesis compound P1 and also the ninhydrin-negative compound S2 to of R. tropici CIAT899, selection the ninhydrin-negative compound P2, presumably by hydroxy- of acid-sensitive mutants and cosmid complementation. lation at a still-unknown position in these molecules. As S1 Tn5 mutagenesis of R. tropici CIAT899 was carried out using and S2 both seem to function as in vivo substrates for the OlsC E. coli S17-1 carrying pSUP1021 as donor strain (Simon et al. reaction, one might expect that S1 and S2 would have similar 1983). Transconjugants carrying the transposon were isolated, chemical structures and that, therefore, S2 might be a modified and acid-sensitive mutants were selected on different acidified version of the OL known to date. The nonreactivity of S2 with media, as previously described (Vinuesa et al. 2003). A cosmid ninhydrin might be due to an additional, so far unknown modi- library of R. tropici CIAT899 made in pVK101 (Vargas et al. fication at the δ-amino group of its ornithine residue. Simi- 1990) was mobilized en masse into CIAT899-PV9, by tripar- larly, as P1 and P2 might be both products of an in vivo reac- ental mating using pRK2013 as helper plasmid (Figurski and tion catalyzed by OlsC, again one can expect that P1 and P2 Helinski 1979). Transconjugants that restore symbiotic profi- would have similar structures, with P2 having an additional ciency on common beans were isolated as previously described modification at the δ-amino group of its ornithine residue. The (Vinuesa et al. 2003). near-complete lack of S1 and S2 in the case of strain 899- olsC∆1/pBBR-1,6BE can be explained by a more efficient Standard DNA manipulations. conversion of S1 and S2 to P1 and P2 due to an increased copy Genomic DNA from rhizobial strains was isolated using the number of the gene responsible for the conversion. GenomicPrep cells and tissues DNA isolation kit (Amersham Mutant 899-olsC∆1 is acid-tolerant but symbiotically defec- Biosciences, Buckinghamshire, U.K.), following the manufac- tive. In contrast, the complemented mutant 899-olsC∆1/pBBR- turer’s instructions. Plasmid DNA from E. coli cultures was 1,6BE was able to restore the symbiotic proficiency but was isolated with the High pure plasmid isolation kit (Roche, acid-sensitive. The latter suggests that lipids P1 and P2 are Mannheim, Germany). Restriction endonucleases were pur- necessary for a successful symbiotic interaction of R. tropici chased from New England Biolabs (Beverly, MN, U.S.A.) and CIAT899 with the plant host, whereas lipids S1 and S2 are re- were used according to standard procedures (Sambrook et al. quired for acid tolerance. This hypothesis is consistent with the 1989). PCR amplifications were carried out in a Gene Amp fact that, in the parent strain, CIAT899 all four distinct classes PCR system 2700 (Applied Biosystems, Foster City, CA, of OL are present and that this strain is both acid-tolerant and U.S.A.), using Taq (Roche) or XL polymerase (Applied Bio- symbiotically proficient. systems) in a standard 50-µl PCR mix as previously described The analysis performed on mutant 899-olsC∆1 permits us to (Vinuesa et al. 1999). report for the first time that membrane lipids of R. tropici are To map the transposon insertion in mutant 899-PV9, total involved in symbiosis and that the putative β-hydroxylase en- DNA of this strain was digested with EcoRI, was transferred coded by olsC is part of a biosynthesis pathway for membrane to a nylon membrane, and was hybridized with a DIG- lipids not previously described for any species. We demon- labeled probe generated by the incorporation of DIG-UTP strate that R. tropici lipid species S1, S2, P1, and P2 are indeed (Roche) into the nptII marker of Tn5 via PCR, as previously OL and that the product encoded by olsC is necessary to con- described (Vinuesa et al. 2003). The single hybridizing frag- vert lipids S1 and S2 to P1 and P2, presumably by hydroxyla- ment (approximately 14 kb) was cloned into pBluescript tion at a still-unknown position. Lipids P1 and P2, which pre- (pSK), yielding p899PV9ESK. This plasmid was used as sumably carry a hydroxyl group at a still-unkown position, are template for PCR amplification with primers Tn5-77/58EB required for an effective symbiotic interaction with bean (Vinuesa et al. 2003) and M13 universal to map the Tn5 plants, while a lack of lipids S1 and S2 was correlated with insertion by sequencing the resulting PCR product with the acid sensitivity. Hydroxylations at the 2-position of fatty acyl former primer, which reads outwardly from the Tn5 insertion residues of membrane lipids such as PE or OL are known to elements. The amplification product was cloned into pSK, occur in Burkholderia cepacia and other bacteria (Taylor et al. yielding p899PV9E-PCR, and was used to generate a hybridi- 1998). Although the molecular functions of OL modifications zation probe of the genomic DNA flanking the Tn5 insertion in R. tropici are not understood at this point, a balanced mem- by DIG-labeling. brane lipid composition of the R. tropici cell membranes is re- Cosmids were isolated and restricted with several enzymes quired for both the symbiotic interaction with plants and for as previously described (Vinuesa et al. 2003). Cosmid acid tolerance under free-living conditions. This is the first re- pPV9cos2 was hybridized with the probe derived from the in- port of a rhizobial membrane lipid other than phospholipids sert cloned in p899PV9E-PCR, and the resulting hybridizing with relevance for symbiosis. band was cloned into pSK, yielding pPV9E-SK. Subclones from pPV9E-SK were sequenced with an ABI Prism 3700 MATERIALS AND METHODS automated sequencer, using the universal M13f and M13r primers (Applied Biosystems). PCR primers were subsequently Bacterial strains and plasmids. designed to obtain an approximately 8× coverage of overlap- Bacterial strains and plasmids used in the present work are ping plasmid subclones and PCR products from which a contig listed in Table 2. Rhizobial strains were grown in PY (Noel was assembled using SeqManII from the DNASTAR package et al. 1984), minimal medium (MM) (Kingsley and Bohlool (Lasergene, Madison, WI, U.S.A.). 1182 / Molecular Plant-Microbe Interactions DNA sequence analyses. derived expression. Plasmid pBRR-1,6BE was transferred into ORF with high coding probability were identified on the mutant 899-olsC∆1 by triparental mating, and transconjugants contig sequence using FrameD, with the S. meliloti codon usage were selected on PY and MM plates amended with Gm, Ap, table and the pentanucleotide aagga as ribosomal binding site. and Nal. The empty pBBR-MCS5 vector was transferred into The BLASTX and BLASTP programs from the nonredundant 899-olsC∆1 as a control of the complementation, and trans- sequence databases at the National Center for Biotechnology conjugants were selected as mentioned before. Information were used to search for homology. Protein sequence analyses to predict secondary structure, cell localization, trans- Plant tests. membrane domains, conserved motifs, and hydrophobicity P. vulgaris seeds were surface-sterilized with 1.2% sodium profiles were performed using Prosite, ProDom, PsortB, hypochlorite and were germinated on 1% agar-water plates as TmPred, PsPred, Pfam, and iProClass program servers. A described (Vinuesa et al. 1999). Seedlings were transferred to search for putative promoter regions in intergenic regions was 250-ml flasks filled with vermiculite and nitrogen-free nutrient performed using the NNPP server. solution (Fahraeus 1957) and were inoculated with about 105 CFU per plant. Plants were grown in a controlled growth Determination of operon structure by RT-PCR. chamber and harvested 21 dpi. Nitrogenase activity of nodu- RNA from R. tropici CIAT899 was isolated using the High lated roots was determined by acetylene reduction assay. Com- pure RNA isolation kit (Roche) according to the manufac- petition experiments were performed by coinoculating the mu- turer’s instructions, and cDNA was immediately synthesized tant strain together with a gusA-tagged reported strain, using the Omniscript RT kit (Qiagen, Hilden, Germany). This CIAT899-G1, at a low-inoculum titer (approximately 500 cells cDNA was used as template for PCR amplification of the IGS per plantlet) to minimize nodule coinfections, as previously between sycA and olsC, using primers PV9-2000f (5′gcagcggc described (Vinuesa et al. 2003). Plants were harvested 21 dpi, cataccagcatc) and PV9-2985r (5′tcacgccgaaaccgaggag). Posi- and blue nodules were counted after β-glucuronidase staining tive controls included the amplification of the 16S rDNA gene (Wilson et al. 1995). with primers fD1 and rD1 (Weisburg et al. 1991) and the am- plification of a 389-bp internal fragment of olsC using primers In vivo labeling of rhizobial strains with [14C]acetate PV9-2571f-H (5′ccaagcttcctcccggaccgcac) and PV9-2960r-B or [14C]ornithine and analysis of lipid extracts. (5′ccggatccagcgggtgtcggtgg). To discard the presence of con- The lipid compositions of R. tropici CIAT899, mutant 899- tamination by R. tropici genomic DNA, the same master mix olsC∆1, and complemented mutant 899-olsC∆1/pBBR1,6BE used for cDNA synthesis but lacking the retrotranscriptase was were determined after labeling with [1-14C]acetate (60 used as a negative control for PCR amplification experiments mCi/mmol; Amersham) during growth on PY medium for 24 with the primers for the 16S rDNA gene and the internal olsC h. The incorporation of ornithine into lipids was followed by fragment . labeling R. tropici CIAT899 with DL-[1-14C]ornithine (56 mCi/mmol; Amersham) during growth on MM for 48 h. Cul- Construction tures (2 ml) were inoculated from precultures to an initial opti- of nonpolar mutants 899-sycA∆1 and 899-olsC∆1. cal density at 600 nm of 0.05 in the respective medium. After To construct a nonpolar mutant in sycA, PCR amplification the addition of 1 µCi [1-14C]acetate or of 0.5 µCi DL-[1- of two nonoverlapping approximately 400-bp fragments lo- 14C]ornithine to the respective cultures, they were incubated at cated at the opposite ends of the ORF (resulting in a 991-bp 28°C with appropriate shaking. At the end of the growth pe- deletion) were performed using primer pairs (restriction sites riod, cells were harvested by centrifugation and resuspended in underlined) PV9-524f-H (5′gcaagcttgcccgggcggtgtgacg) and 100 µl of water, and lipid extracts were obtained according to PV9-971r-B (5′cggatccggcaacgggcataagaaag) and PV9-1962f- Bligh and Dyer (1959). Aliquots of the lipid extracts were B (5’ccggatcctcgctgtcgcgtgctt) and PV9-2272r-E (5′ccgaattcct spotted on high-performance TLC silica gel 60 (Merck, Poole, gccatcggagcgtc) and total DNA of R. tropici as template (Fig. U.K.) plates and were separated in two dimensions using chlo- 1). The same strategy was used to construct a 211-bp deletion roform/methanol/water (140:60:10, vol/vol/vol) as a mobile in olsC, using primer pairs PV9-2571f-H (5′ccaagcttcctcccg phase for the first dimension and chloroform/methanol/ gaccgcac) and PV9-2960r-B (5′ccggatccagcgggtgtcggtgg), and acetic acid (130:50:20, vol/vol/vol) for the second. Primary PV9-3171f-B (5′cggatccgcgtcgacaatcacg) and PV9-3622r-E amine-containing lipids were visualized by spraying the plates (5′cgcgaattcggtggcggcatgacg). The PCR products were digested with a solution of 0.2% ninhydrin in acetone and subsequent with HindIII plus BamHI and BamHI plus EcoRI, respectively, treatment at 100°C for 5 min. To quantify the membrane lipid and were ligated to HindIII plus EcoRI–restricted composition, developed 2D-TLC plates were stained with io- pK18mobsacB. The resulting plasmids pKR∆02 and pKR∆03 dine and the radioactivity of individual spots was quantified in were transferred into strain CIAT899 by triparental matings, a scintillation counter, as previously described (Geiger et al. and double recombinants were selected on PY medium at pH 1999). 6.8 amended with 12% sucrose, as previously reported (Vinuesa et al. 2003). Two nonpolar mutants, hereafter named ACKNOWLEDGMENTS 899-sycA∆1 and 899-olsC∆1, were obtained. We acknowledge the support of the German Science Foundation through Complementation of strain 899-olsC∆1 with pBRR-1,6BE. SFB 395 (Project A6), the European Union for INCO-DEV Project (ICA4- A 1,660-bp EcoRI-BamHI fragment was amplified with CT-2001-10057) and DGAPA-UNAM-Mexico (PAPIIT grants 200802 and primers PV9-1962f-B and PV9-3622r-E and was cloned into 205802) for financial support. pBBR-MCS5 (Gmr), a broad-host range vector (Kovach et al. 1995) to obtain pBRR-1,6BE. The cloned fragment contains LITERATURE CITED the whole gene encoded by olsC and an additional 534 bp Aarons, S. R., and Graham, P. H. 1991. Response of Rhizobium legumino- upstream of the predicted start codon with the region carrying sarum bv. phaseoli to acidity. Plant Soil 134:145-151. the putative promoter sequence predicted by the NNPP server Accardi, A., and Miller, C. 2004. Secondary active transport mediated by a (Fig. 1). This fragment was cloned in the opposite direction to prokaryotic homologue of ClC Cl– channels. Nature 427:803-807. the lacZ promoter of the vector, in order to avoid vector- Accardi, A., Kolmakova-Partensky, L., Williams, C., and Miller, C. 2004. Vol. 18, No. 11, 2005 / 1183 Ionic currents mediated by a prokaryotic homologue of CLC Cl– chan- alpha-ketoglutarate-dependent dioxygenases. Proc. Natl. Acad. Sci. nels. J. Gen. Physiol. 123:109-119. U.S.A. 91:7227-7231. Ballen, K., Graham, P., Jones, R., and Bowers, J. 1990. Acidity and cal- Jjemba, P. K. 2001. The interaction of protozoa with their potential prey cium interaction affecting cell envelope stability in Rhizobium. Can. J. bacteria in the rhizosphere. J. Eukaryot. Microbiol. 48:320-324. Microbiol. 44:582-587. Kingsley, M. T., and Bohlool, B. 1992. Extracellular polysaccharide is not Bligh, E. G., and Dyer, J. W. 1959. A rapid method of total lipid extraction responsible for aluminum tolerance of Rhizobium leguminosarum bv. and purification. Can. J. Biochem. Physiol. 37:911-917. phaseoli CIAT899. Appl. Environ. Microbiol. 58:1095-1101. Chen, H., Richardson, A. E., and Rolfe, B. G. 1993a. Involvement of Kovach, M. E., Elzer, P. H., Hill, D. S., Robertson, G. T., Farris, M. A., genes on a megaplasmid in the acid-tolerant phenotype of Rhizobium Roop, R. M., 2nd, and Peterson, K. M. 1995. Four new derivatives of leguminosarum biovar trifolii. Appl. Environ. Microbiol. 6:1058-1064. the broad-host-range cloning vector pBBR1MCS, carrying different Chen, H., Richardson, A. E., and Rolfe, B. G. 1993b. Studies of the phy- antibiotic-resistance cassettes. Gene 166:175-176. siological and genetic basis of acid tolerance in Rhizobium leguminosa- Lodwig, E., Hosie, A., Bourdes, A., Findlay, K., Allaway, D., Karunakaran, rum biovar trifolii. Appl. Environ. Microbiol. 6:1798-1804. R., Downie, J., and Poole, P. S. 2003. Amino-acid cycling drives nitrogen Chen, T. Y. 2005. Structure and function of ClC channels. Annu. Rev. fixation in the legume-Rhizobium symbiosis. Nature 422:722-726. Physiol. 67:809-839. López-Lara, I. M., Sohlenkamp, C., and Geiger, O. 2003. Membrane lipids Day, D. A., Poole, P. S., Tyerman, S. D., and Rosendahl, L. 2001. Ammo- in plant-associated bacteria: Their biosyntheses and possible functions. nia and amino acid transport across symbiotic membranes in nitrogen- Mol. Plant-Microbe Interact. 16:567-579. fixing legume nodules. Cell. Mol. Life Sci. 58:61-71. Marschner, H. 1995. Mineral Nutrition of Higher Plants. Academic Press, de Rudder, K. E. E., Thomas-Oates, J. E., and Geiger, O. 1997. Rhizobium London. meliloti mutants deficient in phospholipid N-methyltransferase still con- Martínez-Romero, E., Segovia, L., Mercante, F. M., Franco, A. A., Graham, tain phosphatidylcholine. J. Bacteriol. 179:6921-6928. P., and Pardo, M. A. 1991. Rhizobium tropici, a novel species nodulat- Ding, Y., and Waldor, M. K. 2003. Deletion of a Vibrio cholerae ClC chan- ing Phaseolus vulgaris L. beans and Leucaena sp. trees. Int. J. Syst. nel results in acid sensitivity and enhanced intestinal colonization. Bacteriol. 41:417-26. Infect. Immun. 71:4197-4200. Merrell, D. S., and Camilli, A. 2002. Acid tolerance of gastrointestinal Ditta, G., Stanfiels, S., Corbin, D., and Helinski, D. 1980. Broad host pathogens. Curr. Opin. Microbiol. 5:51-55. range DNA cloning system from gram-negative bacteria: Construction Minder, A. C., de Rudder, K. E., Narberhaus, F., Fischer, H. M., Hennecke, of a gene bank of Rhizobium meliloti. Proc. Natl. Acad. Sci. U.S.A H., and Geiger, O. 2001. Phosphatidylcholine levels in Bradyrhizobium 77:7347-7351. japonicum membranes are critical for an efficient symbiosis with the Dutzler, R. 2004. Structural basis for ion conduction and gating in ClC soybean host plant. Mol. Microbiol. 39:1186-1198. chloride channels. FEBS (Fed. Eur. Biochem. Soc.) Lett. 564:229-233. Munns, D., Hohenberg, T., Righetti, T., and Lauter, D. 1981. Soil acidity Dutzler, R., Campbell, E. B., Cadene, M., Chait, B. T., and MacKinnon, R. of symbiotic and nitrogen-fertilized soybeans. Agron. J. 73:407-410. 2002. X-ray structure of a ClC chloride channel at 3.0 Å reveals the mo- Munns, D., Keyser, H., Fogle, V., Hohenberg, J., Righetti, T., Lauter, D., lecular basis of anion selectivity. Nature 415:287-294. Zaroug, M., Clarkin, K., and Whitacre, K. 1979. Tolerance of soil acid- Dutzler, R., Campbell, E. B., and MacKinnon, R. 2003. Gating the selec- ity in symbioses of mung bean with rhizobia. Agron. J. 71:256-260. tivity filter in ClC chloride channels. Science 300:108-112. Noel, K. D., Sánchez, A., Fernández, L., Leemans, J., and Cevallos, M. A. Fahraeus, G. 1957. The infection of clover root hair by nodule bacteria 1984. Rhizobium phaseoli symbiotic mutants with transposon Tn5 in- studied by a single glass slide technique. J. Gen. Microbiol. 16:374- sertions. J. Bacteriol. 158:148-155. 381. Nogales, J., Campos, R., BenAbdelkhalek, H., Olivares, J., Lluch, C., and Figurski, D. H., and Helinski, D. R. 1979. Replication of an origin-con- Sanjuan, J. 2002. Rhizobium tropici genes involved in free-living salt taining derivative of plasmid RK2 dependent on a plasmid function pro- tolerance are required for the establishment of efficient nitrogen-fixing vided in trans. Proc. Natl. Acad. Sci. U.S.A. 76:1648-1652. symbiosis with Phaseolus vulgaris. Mol. Plant-Microbe Interact. Foster, J. W. 1999. When protons attack: Microbial strategies of acid adap- 15:225-232. tation. Curr. Opin. Microbiol. 2:170-174. O’Hara, G. W., and Glenn, A. R. 1994. The adaptive acid tolerance re- Gao, J. L., Weissenmayer, B., Taylor, A. M., Thomas-Oates, J., López- sponse in root nodule bacteria and Escherichia coli. Arch. Microbiol. Lara, I. M., and Geiger, O. 2004. Identification of a gene required for 161:286-292. the formation of lyso-ornithine lipid, an intermediate in the biosynthesis Peick, B., Graumann, P., Schmid, R., Marahhiel, M., and Werner, D. 1999. of ornithine-containing lipids. Mol. Microbiol. 53:1757-1770. Differential pH-induced proteins in Rhizobium tropici CIAT899 and Geiger, O., Röhrs, V., Weissenmayer, B., Finan, T. M., and Thomas-Oates, Rhizobium etli CIAT611. Soil Biol. Biochem. 31:189-194. J. E. 1999. The regulator gene phoB mediates phosphate stress-con- Reeve, W. G., Dilworth, M. J., Tiwari, R. P., and Glenn, A. R. 1997. Regu- trolled synthesis of the membrane lipid diacylglyceryl-N,N,N-trimethyl- lation of exopolysaccharide production in Rhizobium leguminosarum homoserine in Rhizobium (Sinorhizobium) meliloti. Mol. Microbiol. biovar viciae WSM710 involves exoR. Microbiology 143:1951-1958. 32:63-73. Riccillo, P. M., Muglia, C. I., de Bruijn, F. J., Roe, A. J., Booth, I. R., and Gibbons, H. S., Lin, S., Cotter, R. J., and Raetz, C. R. 2000. Oxygen re- Aguilar, O. M. 2000. Glutathione is involved in environmental stress re- quirement for the biosynthesis of the S-2-hydroxymyristate moiety in sponses in Rhizobium tropici, including acid tolerance. J. Bacteriol. Salmonella typhimurium lipid A. Function of LpxO, a new Fe2+/alpha- 182:1748-1753. ketoglutarate-dependent dioxygenase homologue. J. Biol. Chem. Sambrook, J., Fritsch, E. F., and Maniatis, T. 1989. Molecular Cloning: A 275:32940-32949. Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Glenn, A. R., Reeve, W. G., Tiwari, R. P., and Dilworth, M. J. 1999. Acid Harbor, NY, U.S.A. tolerance in root nodule bacteria. Novart. Fdn. Symp. 221:112-126. Schäfer, A., Tauch, A., Jäger, W., Kalinowski, G., Thierbach, G., and Graham, P. H. 1992. Stress tolerance in Rhizobium and Bradyrhizobium, Pühler, A. 1994. Small mobilizable multi-purpose cloning vectors de- and nodulation under adverse soil conditions. Can. J. Microbiol. rived from the Escherichia coli plasmids pK18 and pK19: Selection of 38:475-484. defined deletions in the chromosome of Corynebacterium glutamicum. Graham, P. H., Viteri, S. E., Mackie, F., Vargas, A. A. T., and Palacios, A. Gene 145:69-73. 1982. Variation in acid soil tolerance among strains of Rhizobium Simon, R., Priefer, U., and Pühler, A. 1983. A broad host range mobiliza- phaseoli. Field Crop Res. 5:121-128. tion system for in vivo genetic engineering: Transposon mutagenesis in Graham, P. H., Draeger, K. J., Ferrey, M. L., Conroy, M. J., Hammer, B. gram-negative bacteria. Bio/Technology 1:784-791. E., Martínez, E., Aarons, S. R., and Quinto, C. 1994. Acid pH tolerance Sohlenkamp, C., López-Lara, I. M., and Geiger, O. 2003. Biosynthesis of in strains of Rhizobium and Bradyrhizobium, and initial studies on the phosphatidylcholine in bacteria. Prog. Lipid Res. 42:115-162. basis for acid tolerance of Rhizobium tropici UMR1899. Can. J. Taylor, C. J., Anderson, A. J., and Wilkinson, S. G. 1998. Phenotypic vari- Microbiol. 40:189-207. ation of lipid composition in Burkholderia cepacia: A response to Hungria, M., and Vargas, M. 2000. Environmental factors affecting N2 increased growth temperature is a greater content of 2-hydroxy acids in fixation in grain legumes in the tropics, with an emphasis on Brazil. phosphatidylethanolamine and ornithine amide lipid. Microbiology Field Crop Res. 65:151-164. 144:1737-1745. Iyer, R., Iverson, T. M., Accardi, A., and Miller, C. 2002. A biological role Udvardi, M. K., and Day, D. A. 1997. Metabolite transport across symbi- for prokaryotic ClC chloride channels. Nature 419:715-718. otic membranes of legume nodules. Annu. Rev. Plant Physiol. Plant Jia, S., McGinnis, K., VanDusen, W. J., Burke, C. J., Kuo, A., Griffin, P. Mol. Biol. 48:493-523. R., Sardana, M. K., Elliston, K. O., Stern, A. M., and Friedman, P. A. Vargas, C., Martínez, J. L., Megías, M., and Quinto, C. 1990. Identifica- 1994. A fully active catalytic domain of bovine aspartyl (asparaginyl) tion and cloning of nodulation genes and host specificity determinants beta-hydroxylase expressed in Escherichia coli: Characterization and of the broad host-range Rhizobium leguminosarum biovar phaseoli evidence for the identification of an active-site region in vertebrate strain CIAT899. Mol. Microbiol. 11:1899-1910. 1184 / Molecular Plant-Microbe Interactions Vinuesa, P., Reuhs, B. L., Breton, C., and Werner, D. 1999. Identification Wilson, K. J., Sessitsch, A., Corbo, J. C., Giller, K. E., Akkermans, A. D., of a plasmid-borne locus in Rhizobium etli KIM5s involved in lipopoly- and Jefferson, R. A. 1995. Beta-Glucuronidase (GUS) transposons for saccharide O-chain biosynthesis and nodulation of Phaseolus vulgaris. ecological and genetic studies of rhizobia and other gram-negative bac- J. Bacteriol. 181:5606-5614. teria. Microbiology 141:1691-1705. Vinuesa, P., Neumann-Silkow, F., Pacios-Bras, C., Spaink, H. P., Martínez- Zahran, H. H. 1999. Rhizobium-legume symbiosis and nitrogen fixation Romero, E., and Werner, D. 2003. Genetic analysis of a pH-regulated under severe conditions and in an arid climate. Microbiol. Mol. Biol. operon from Rhizobium tropici CIAT899 involved in acid tolerance and Rev. 63:968-989. nodulation competitiveness. Mol. Plant-Microbe Interact. 16:159-168. Zhu, X., and Williamson, P. R. 2003. A CLC-type chloride channel gene is Walker, T. S., Bais, H. P., Grotewold, E., and Vivanco, J. M. 2003. Root required for laccase activity and virulence in Cryptococcus neoformans. exudation and rhizosphere biology. Plant Physiol. 132: 44-51. Mol. Microbiol. 50:1271-1281. Weisburg, W. G., Barns, S. M., Pelletier, D. A., and Lane, D. J. 1991. 16S ribosomal DNA amplification for phylogenetic study. J. Bacteriol. 173:697-703. AUTHOR-RECOMMENDED INTERNET RESOURCES Weissenmayer, B., Gao, J. L., López-Lara, I. M., and Geiger, O. 2002. Identification of a gene required for the biosynthesis of ornithine- Genopole’s FrameD gene finder program: derived lipids. Mol. Microbiol. 45:721-733. genopole.toulouse.inra.fr/bioinfo/FrameD/FD Werner, D., Wilcockson, J., and Zimmermann, E. 1975. Adsorption and se- The Berkeley Drosophila Genome Project NNPP server: lection of rhizobia with ion-exchange papers. Arch. Microbiol. 105:27-32. www.fruitfly.org/seq_tools/promoter.html Vol. 18, No. 11, 2005 / 1185