A purL mutant of Sinorhizobium fredii HH103 is symbiotically defective and altered in its lipopolysaccharide

Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 ApurL mutant of Sinorhizobium fredii HH103 is symbiotically defective and altered in its lipopolysaccharide Ana M. Buendı´a-Claverı´a, 1 Ahmed Moussaid, 1 F. Javier Ollero, 1 Jose ´M. Vinardell, 1 Antonio Torres, 1 Javier Moreno, 2 Antonio M. Gil-Serrano, 3 Miguel A. Rodrı´guez-Carvajal, 3 Pilar Tejero-Mateo, 3 Jan L. Peart, 4 Nicholas J. Brewin 4 and Jose ´E. Ruiz-Sainz 1 Correspondence Jose ´E. Ruiz-Sainz [email protected] 1 Departamento de Microbiologı´a, Facultad de Biologı´a, Universidad de Sevilla, Apdo 1095, 41080 Sevilla, Spain 2 Departamento de Biologı´a Celular, Facultad de Biologı´a, Universidad de Sevilla, Av. Reina Mercedes s/n, 41012 Sevilla, Spain 3 Departamento de Quı´mica Orga ´nica, Facultad de Quı´mica, Universidad de Sevilla, Apdo 553, 41071 Sevilla, Spain 4 Department of Molecular Microbiology, John Innes Centre, Norwich NR4 7UH, UK Received 5 November 2002 Revised 13 March 2003 Accepted 18 March 2003 The pleiotropic phenotype of an auxotrophic purL mutant (SVQ295) of Sinorhizobium fredii HH103 has been investigated. SVQ295 forms colonies that are translucent, produce more slime and absorb less Congo red than those of wild-type strain HH103. SVQ295 did not grow in minimal medium unless the culture was supplemented with thiamin and adenine or with thiamin and AICA-riboside (5-aminoimidazole-4-carboxamide 1-b-D-ribofuranoside), an intermediate of purine biosynthesis. Bacterial cultures supplemented with AICA-riboside or adenine reached the same culture density, although the doubling time of SVQ295 cultures containing AICA-riboside was clearly longer. S. fredii SVQ295 induced pseudonodules on Glycine max and failed to nodulate six different legumes. On Glycyrrhiza uralensis, however, nodules showing nitrogenase activity and containing infected plant cells were formed. SVQ295 showed auto-agglutination when grown in liquid TY medium and its lipopolysaccharide (LPS) electrophoretic profile differed from that of its parental strain HH103-1. In addition, four monoclonal antibodies that recognize the LPS of S. fredii HH103 failed to recognize the LPS produced by SVQ295. In contrast, 1 H-NMR spectra of K-antigen capsular polysaccharides (KPS) produced by SVQ295 and the wild-type strain HH103 were similar. Co-inoculation of soybean plants with SVQ295 and SVQ116 (a nodA mutant derivative of HH103) produced nitrogen-fixing nodules that were only occupied by SVQ116. INTRODUCTION The interaction between rhizobia and legumes to form nitrogen-fixing root nodules involves several communication steps to coordinate gene expression and nodule development. The initial exchange of signals activates rhizobial nod gene expression by plant-produced flavonoid compounds. The nod gene products are responsible for the production of host-specific lipooligosaccharide signal molecules that cause root hair curling and nodule meristem initiation (Fisher & Long, 1992; Day et al., 2000). Unlike the bacterial signals specified by nod genes, signal molecules that may be required to initiate and maintain infection thread development are poorly understood. Studies with rhizobial mutants defective in exopolysaccharide, lipopolysaccharide (LPS) or K-antigen capsular polysaccharides (KPS) indicate that these polysaccharides are important for infection of a variety of legumes (Becker &Pu ¨hler, 1998; Kannenberg et al., 1998; Kannenberg & Carlson, 2001). Interestingly, Sinorhizobium fredii and S. meliloti produce structurally conserved LPS but strainspecific KPS, indicating that the conserved LPS lack the Abbreviations: AICA-riboside, 5-aminoimidazole-4-carboxamide 1-b-Dribofuranoside; AICAR, 5-aminoimidazole-4-carboxamide ribonucleotide; KPS, K-antigen capsular polysaccharide; LCO, lipo-chitin oligosaccharide. The GenBank accession number for the sequence determined in this work is AF275718. 0002-6099 G2003 SGM Printed in Great Britain 1807 Microbiology (2003), 149, 1807–1818 DOI 10.1099/mic.0.26099-0 Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 structural information necessary to influence host specificity while the variable KPS may affect strain–cultivar interactions (Reuhs et al., 1998). Moreover, studies using a panel of anti-S. meliloti monoclonal antibodies that recognizes LPS or KPS have revealed changes of these polysaccharides during symbiosis (Reuhs et al., 1999). The phenotypes of different Rhizobium auxotrophs have also provided interesting clues to the physiology of nodule development. By analysing the symbiotic behaviour of auxotrophic mutants, it has been demonstrated that some metabolic pathways of Rhizobium bacteria are important for an efficient symbiotic interaction (Pain 1979; Noel et al., 1988). In some legume–Rhizobium associations, the host plant can supply the auxotrophic symbiont with the required nutrient(s) (Djordjevic et al., 1988). However, the success of this physiological complementation will depend on the host plant, the rhizobial strain, and the nature of the auxotrophy (Beringer et al., 1980). One intriguing type of finding is that in some biosynthetic pathways, intermediates, rather than endproducts, may be required for the bacteria to elicit nodule development or infection. For example, studies with S. meliloti tryptophan auxotrophs suggest that anthranilate synthesis is important for bacteroid development symbiosis with alfalfa (Barsomian et al., 1992). One of the most universally deleterious symbiotic defects is purine auxotrophy. Purine auxotrophs of most rhizobial species are unable to nodulate their host plants effectively (Pankhurst & Schwinghamer, 1974; Denarie ´et al., 1976; Pain, 1979). S. meliloti purine auxotrophs have been reported to induce ineffective nodules on alfalfa (Denarie ´et al., 1976). Several purine auxotrophs of Rhizobium leguminosarum bv. viciae were described as noninfective or even being unable to induce nodule formation (Pain, 1979). Similarly, a purine auxotroph of the broad-host-range Rhizobium sp. NGR234 elicited root hair curling and nodule meristem initiation on Macroptilium atropurpureum, but no infection threads were observed (Djordjevic et al., 1988). On soybean, S. fredii purine auxotrophs induced pseudonodules that did not contain bacteria (Kim et al., 1988). Purine auxotrophs (Pur 2 )ofR. etli CFN42 elicited pseudonodules on bean plants (Noel et al., 1988). These mutants caused root hair curling and nodule meristem initiation but did not elicit infection threads. Supplementation of the root medium with 0?1 mM purine, or purine nucleosides, did not enhance nodulation ability of R. etli pur mutants. However, the addition of 0?1 mM 5-aminoimidazole-4carboxamide 1-b-D-ribofuranoside (AICA-riboside, an intermediate in the biosynthetic purine pathway) to the plant nutritive solution significantly enhanced nodule development (Noel et al., 1988; Newman et al., 1994). Nodules induced by the mutants in the presence of AICAriboside resembled those formed by the wild-type, but were unpigmented and lacked nitrogenase activity. Examination of these nodules by light microscopy revealed the presence of infection threads filled with bacteria, but infected plant cells were not observed. Time-course experiments showed that the addition of AICA-riboside to Phaseolus vulgaris roots within the first 6 days after inoculation with a R. etli pur mutant allowed the formation of vascular bundles in the nodule cortex. Beyond this time, AICA-riboside supplementation produced nodules that contained their vascular bundles in a central position (Newman et al., 1992). AICAriboside also promotes Rhizobium leguminosarum bv. viciae and Sinorhizobium fredii pur mutants for the formation of root nodules in Pisum sativum and Glycine max, respectively (Newman et al., 1994). All these results have led to the hypothesis that rhizobia might convert AICA riboside into AICAR (5-aminoimidazole-4-carboxamide ribonucleotide) to initiate and/or sustain infection thread development, possibly by using this substance in the production of a signal molecule. We report here the isolation and characterization of a purL mutant (SVQ295) of S. fredii HH103 that requires the addition of adenine and thiamin to grow on minimal medium and forms pseudonodules on Glycine max roots. We demonstrate that the symbiotic phenotype of SVQ295 varies depending on the legume assayed. In Macrotyloma axyllare no nodules were formed, while in Glycyrrhiza uralensis, nodules containing infected cells were observed. Furthermore, the LPS profile of mutant SVQ295 is different from that shown by its wild-type parent HH103, and monoclonal antibodies raised against the HH103 LPS failed to recognize the LPS of SVQ295. METHODS Strains, plasmids and media. The bacterial strains and plasmids used in this work are listed in Table 1. Sinorhizobium strains were grown in TY medium (Beringer, 1974), yeast mannitol (YM) medium (Vincent, 1970), B 2 medium (Spaink et al., 1992) or MMB (Bergensen, 1961) at 28 uC. To screen for auxotrophic strains the minimal medium (MMB) was solidified with agarose at 1?5 % (w/v). Escherichia coli was cultured in Luria–Bertani (LB) medium (Maniatis et al., 1982) at 37 uC. When required, the media were supplemented with the appropriate antibiotics as described by Lamrabet et al. (1999). When required, MMB medium was supplemented with adenine (12 mgml 21 ), AICA-riboside (0?1or0?5 mM) and/or thiamin (0?8mgml 21 ). Genetic manipulations. Plasmids were transferred by conjugation as described by Simon (1984). Random transposon Tn5-Mob mutagenesis of Sinorhizobium fredii strain HH103-1 was carried out by using the suicide plasmid pSUP5011 as described by Simon (1984). Transconjugants were selected on TY medium containing kanamycin and streptomycin. Single colonies with altered morphology on solid TY or YM media were tested for auxotrophy by using a modification of the Holliday test (Beringer et al., 1984). Total genomic DNA and plasmid isolations were carried out as described by Maniatis et al. (1982). DNA manipulation, including restriction analyses, ligation, bacterial transformation and electrophoresis were performed according to the general protocols of Maniatis et al. (1982). The suicide rhizobial plasmid pSUP202 (Simon et al., 1983) was used to clone the rhizobial DNA that is adjacent to the Tn5-Mob insertion in mutant S. fredii SVQ295. Briefly, plasmid pSUP202 was transferred from E. coli S17-1 to mutant SVQ295, and Tc R transconjugants were 1808 Microbiology 149 A. M. Buendı´a-Claverı´a and others Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 selected. Since this plasmid cannot replicate in S. fredii, the only possible inheritance of pSUP202 is by homologous recombination between the Mob site of transposon Tn5-Mob and the Mob region of plasmid pSUP202. This recombination will lead to the integration of plasmid pSUP202 inside transposon Tn5-Mob. Plasmid pSUP202 has a unique EcoRI site (inside the Cm R gene) and this enzyme does not cut the Tn5-Mob transposon. Hence, EcoRI digestion of total DNA isolated from SVQ295 : : pSUP202 generates an EcoRI fragment that contains the tetracycline-resistance gene, the OriV of pSUP202, the Mob site, part of the transposon including the IS50R and a fragment of rhizobial DNA adjacent to the inserted transposon. EcoRI-digested SVQ295 : : pSUP202 total DNA was subjected to self-ligation and E. coli HB101 Tc R transformants were selected. An oligonucleotide (pdad2: 59-AGATTTAGCCCAGTCG-39) was designed from the IS50R DNA sequence of transposon Tn5-Mob and used as a primer to sequence the rhizobial DNA that is adjacent to the right arm of Tn5-Mob in mutant strain SVQ295. DNA sequence was carried out by the dideoxy chain-termination method (Sanger et al., 1977). Computer analysis of the sequences was performed by using the GCG program (University of Wisconsin; Devereux et al., 1984). Plant assays. The nodulation ability of strain SVQ295 was tested on Glycine max (L.) Merr. cv. Williams, Cajanus cajan (L.) Millsp., Macroptilium atropurpureum (Moc. & Sesse ´ex DC) Urb., Vigna radiata (L.) Wilczek, Indigofera tinctoria (L.), Macrotyloma axillare (E. Mey.) Verdc., Albizia lebbek (L.) Benth., Kummerowia stipulacea (Maxim.) Makino, Neonotonia wightii (Arn.) Lackey and Glycyrrhiza uralensis (L.) as described by Buendı ´a-Claverı ´aet al. (1989). Germinated seeds were transferred to Leonard jars containing sterilized vermiculite supplemented with Fa ˚hraeus nutrient solution (Vincent, 1970). When required, the Fa ˚hraeus plant nutritive solution was supplemented with adenine at 12 mgml 21 , thiamin at 0?8mgml 21 and AICA-riboside at 0?1or0?5 mM. Plants were grown for at least 5 weeks with a 16 h photoperiod at 25 uC in the light and 18 uC in the dark. Nitrogenase activity of soybean (Glycine max) nodules was determined by acetylene reduction assays as previously reported (Buendı ´a-Claverı ´aet al., 1986). Plant tops were dried at 70 uC for 48 h and weighed. Bacteria were isolated from surface-sterilized nodules as described by Buendı ´a-Claverı ´a et al. (1986). In double-inoculation experiments, soybean plants cv. Williams were inoculated with 3610 8 bacteria per plant of each co-inoculant, SVQ116 (Rif r Km r , prototrophic) and SVQ295 (Str r Km r , Ade 2 Thi 2 ). Plants were harvested 82 days after co-inoculation. To determinate nodule occupancy, isolates from soybean nodules were tested for antibiotic resistance and auxotrophic markers. Radiolabelling of lipo-chitin oligosaccharide (LCO). Bacteria were grown in 5 ml B 2 medium in the presence of the inducer naringenin (3?6mM) and N-acetyl[ 14 C]glucosamine. The 14 C-labelled LCOs were analysed by TLC as described by Spaink et al. (1992). SDS-PAGE analysis of lipopolysaccharide. Bacterial cultures were grown on solid TY medium or on solid MMB medium supplemented with adenine (or AICA-riboside) and thiamin. Bacterial cells were washed in 0?9 % NaCl and pelleted by centrifugation. The bacterial pellet was resuspended and lysed by heating at 100 uCin 125 ml 60 mM Tris/HCl (pH 6?8) containing 2 % (w/v) SDS and 1 mM EDTA for 5 min and then diluted to 1 ml with the same buffer without SDS. The bacterial crude extract was treated with RNase, DNase and proteinase K as described by Ko ¨pling et al. (1993). The electrophoresis of crude bacterial extracts was performed on a 16?5 % (w/v) polyacrylamide gel with the Tricine buffer system described by Lesse et al. (1990). For the detection of LPS, gels were stained with silver as described by Kittelberger & Hilbink (1993). Monoclonal antibodies (mAbs) and immunoblotting. Production of mAbs against free-living cultures of HH103 or Table 1. Bacterial strains and plasmids Strain or plasmid Derivation and relevant properties Source or reference Sinorhizobium fredii HH103-1 HH103 Str R Buendı ´a-Claverı ´aet al. (1989) SVQ295 HH103-1 purL ::Tn5-Mob This work SVQ269 HH103 Rif R Madinabeitia et al. (2002) SVQ526 Purine auxotrophic mutant of SVQ269 by Tn5-Mob mutagenesis This work SVQ116 SVQ269 nodA ::Tn5-B20 Buendı ´a-Claverı ´aet al. (1996) Rhizobium etli CE3 Derivative of CFN42, Str R Noel et al. (1984) CE382 CE3 purL ::Tn5Newman et al. (1995) Escherichia coli HB101 Restriction-minus, recA background, Str R Boyer & Roulland-Dussoix (1969) Plasmids pBluescript Cloning vector, Ap R Stratagene pSUP202 pBR325 containing the Mob region subcloned in Sau3A, Ap R Cm R Tc R Simon et al. (1983) pSUP5011 Suicide plasmid carrying the Tn5-Mob transposon Simon (1984) pMUS375 pSUP202 containing a 11 kb EcoRI fragment of SVQ295 DNA adjacent to the IS50R from Tn5-Mob This work pMUS509 Cosmid pLAFR1 carrying purL of S. fredii HH103 This work pCOS110 Cosmid pLAFR1 carrying purY,purQ and purL of R. etli CE3 Newman et al. (1994) pJN382A pCOS110 purL ::Tn5Newman et al. (1994) http://mic.sgmjournals.org 1809 purL mutant of S. fredii Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 nodule-derived bacteroids followed previously published procedures (Lucas et al., 1996). Spleen cells derived from immunized male rats (Line LOU/IAP) were fused to the myeloma line IR983F. For immunoblotting experiments, mAbs that recognize HH103 LPS were derived from four culture supernatants. NB3-4.72 and NB3-4.79 (class IgG 2c) were two clones derived from the same hybridoma line originating from a rat immunized with a preparation of HH103 bacteroids, NB6-228.22 (Class IgG 2b) was also obtained following immunization with a bacteroid preparation, while NB7-242.33 (Class IgG 2c) was obtained following immunization with free-living cultures of HH103. Polyacrylamide gels were electroblotted onto nitrocellulose sheets, using a Bio-Rad Semi-dry transblot apparatus at 20 V for 1 h at room temperature. Transfer buffer was 48 mM Tris/HCl, 39 mM glycine, containing 20 % (v/v) methanol and 0?0375 % (w/v) SDS (pH 8?0). Western-blotted sheets were immunostained with the appropriate mAb, using as the secondary antibody a goat anti-rat IgG conjugated to alkaline phosphatase. NMR analyses of KPS. KPS were isolated from S. fredii HH103 and its mutant SVQ295 as described by Gil-Serrano et al. (1999). The samples containing KPS were deuterium-exchanged several times by freeze-drying from 2 H 2 O and then examined in solution (4 mg ml 21 )in99?98 % 2 H 2 O. Spectra were recorded at 303 K on a Bruker AMX 500 spectrometer operating at 500?13 MHz. Chemical shifts are given in p.p.m., using the 2 H 2 O signal (4?75 p.p.m.) as reference. Microscopic studies. Small fragments of nodules and pseudonodules were fixed in 4 % (v/v) glutaraldehyde prepared in 0?1M cacodylate buffer, pH 7?2, for 3 h at 4 uC and post-fixed in 1 % OsO 4 for 2 h at 4 uC. Samples were dehydrated in an acetone series and embedded in Epon (epoxy embedding medium). Toluidineblue-stained semi-thin sections (0?5mm thick) were viewed in a Leitz (Aristoplan) light microscope. RESULTS Isolation of Tn5-Mob mutants showing altered colony morphology S. fredii HH103-1 was mutagenized with Tn5-Mob (Simon, 1984) and 80 000 Nm R transconjugants were screened for alterations in colony morphology. One mutant, called SVQ295, was further investigated. SVQ295 harbours only one copy of Tn5-Mob, as indicated by hybridization experiments using a 0?9kbPstI internal fragment of the kanamycin-resistance gene of Tn5-Mob as a probe (data not shown). It forms colonies that are translucent, produce more slime and absorb less Congo red than those of wildtype strain HH103-1. SVQ295 auto-agglutinates when grown in liquid TY, an indication that this mutant might be altered in its LPS, as reported for R.leguminosarum LPS mutants (Priefer, 1989). PAGE showed that the LPS I and LPS II regions of SVQ295 cultures grown on solid TY medium contain bands that migrate faster than those of its parental strain HH103-1 (Fig. 1A, lanes 2 and 1 respectively). On the other hand, the 1 H-NMR spectra obtained for the KPS from S. fredii HH103 and from its mutant derivative SVQ295 appear to be very similar (Fig. 2). The LCO and plasmid profiles of mutant strain SVQ295 were also identical to those of its parental strain HH103-1 (data not shown). Mutant SVQ295 does not grow in minimal medium (MMB) unless the bacterial culture is supplemented with thiamin and adenine or with thiamin and AICA-riboside, an intermediate of purine biosynthesis. Although the doubling time of SVQ295 cultures grown in Fig. 1. LPS profiles of strains grown in TY medium, unless otherwise indicated. (A) Lane 1, HH103-1; lane 2, SVQ295; lane 3, SVQ295(pMUS509); lane 4, SVQ295(pJN382A); lane 5, SVQ295; lane 6, SVQ295(pCOS110). (B) Lane 1, HH103-1; lane 2, SVQ295; lane 3, SVQ295 (MMB medium supplemented with 0?5 mM AICA-riboside and thiamin); lane 4, SVQ295 (MMB medium supplemented with 0?1 mM AICA-riboside and thiamin); lane 5, SVQ295 (MMB medium supplemented with adenine and thiamin); lane 6, SVQ295 (TY medium supplemented with 0?5 mM AICA-riboside); lane 7, SVQ295 (TY medium supplemented with 0?1 mM AICA-riboside). (C) Lane 1, SVQ295; lane 2, SVQ526; lane 3, HH103-1. 1810 Microbiology 149 A. M. Buendı´a-Claverı´a and others Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 the presence of AICA-riboside is longer (approx. 9 h) than that in the presence of adenine (approx. 5 h), both bacterial cultures (with AICA-riboside or adenine) reached the same OD 660 after 6 days of incubation (5610 8 bacteria ml 21 ). Inoculation of MMB medium supplemented with adenine or AICA-riboside with SVQ295 previously grown in the presence of AICA-riboside showed again that bacteria grew faster in the presence of adenine than in the presence of AICA-riboside (data not shown). Mutant SVQ295 and its parental strain HH103-1 showed the same doubling time (135 min) in complete TY medium. The presence of adenine or AICA-riboside in SVQ295 cultures did not restore the wild-type LPS electrophoretic profile (Fig. 1B). Identification of the mutated gene in SVQ295 Plasmid pSUP202 was used to clone the SVQ295 DNA region that is adjacent to the IS50R of transposon Tn5-Mob. Following the strategy described in Methods, a plasmid (pMUS375) that contained about 16 kb of the SVQ295 DNA that is adjacent to the IS50R of transposon Tn5-Mob was isolated. A XhoI fragment of pMUS375 that contains 485 bp of the IS50R and about 4 kb of SVQ295 DNA was subcloned into plasmid pBluescript. By using primer pdad2, a DNA fragment containing the last 120 bp of the IS50R and 500 bp of the adjacent rhizobial DNA was sequenced. Computer analysis of this sequence indicates homology with the purL gene of S. meliloti, which encodes a 59-phosphoribosylformylglycinamidine synthase (FGAR amidotransferase, EC 6.3.5.3). This enzyme catalyses the step from FGAR (59-phosphoribosyl-N-formylglycinamide) to FGAM (59-phosphoribosyl-N-formylglycinamidine) in the biosynthetic purine pathway and it is essential for biosynthesis of both adenine and thiamin. These results are in accordance with the auxotrophic phenotype exhibited by mutant SVQ295. Symbiotic properties of mutant SVQ295 SVQ295 induces Fix 2 pseudonodules on soybean cv. Williams. Microscopic analysis showed that plant cells of these pseudonodules were devoid of bacteria. Nodulation assays were carried out to investigate whether the addition of thiamin and/or adenine or AICA-riboside to the plant nutrient solution restored soybean nodulation ability of mutant SVQ295. Adenine, thiamin, or both together, did not enhance the symbiotic properties of mutant strain SVQ295 whereas 0?1or0?5 mM AICA-riboside enabled SVQ295 to form infected nodules that did not fix nitrogen and were smaller than those formed by HH103. Bacteria isolated from these nodules had the appropriate antibiotic resistance (Str R Nm R ) and were auxotrophic for adenine and thiamin. The symbiotic properties of SVQ295 were also studied in other leguminous plants that establish nitrogen-fixing symbioses with the wild-type strain HH103-1. Mutant SVQ295 failed to nodulate Cajanus cajan,Indigofera tinctoria,Macrotyloma axillare,Albizia lebbek,Kummerowia stipulacea and Neonotonia wightii. It formed small ineffective nodules on Macroptilium atropurpureum. Although infected nodule cells were observed (Fig. 3D, E), bacteria were not recovered from these nodules. In contrast, SVQ295 was able to form nitrogen-fixing nodules on Glycyrrhiza uralensis (Fig. 3A) as demonstrated by acetylene reduction assays. G. uralensis nodules contained infected plant cells with a large centrally positioned vacuole (Fig. 3B, C) and bacteria isolated from these nodules showed the markers of SVQ295. Isolation and characterization of the purL gene of S. fredii HH103-1 For unknown reasons, it is often difficult to isolate cosmid clones that complement HH103 mutants by conjugational transfer of the HH103 genomic library to the recipient mutant. Because of this difficulty, an HH103 gene library, constructed in cosmid pLAFR1, was transferred to R. etli CE382, a derivative of CE3 containing Tn5in purL (Newman et al., 1995). CE382 Tc R Km R transconjugants were selected on MMB agar and a cosmid named pMUS509 was isolated from one of them. Cosmid pMUS509 was transferred (by transformation) to E. coli S17-1 and from this donor strain it was reintroduced (by conjugation) into R. etli CE382. Transconjugants carrying pMUS509 did not require the presence of adenine and thiamin to grow on MMB medium and were able to form nitrogen-fixing nodules on Phaseolus vulgaris. 54321 d (p.p.m.) Fig. 2. 1 H-NMR spectra (500 MHz) obtained for the capsular polysaccharide from S. fredii HH103 (A) and SVQ295 (B). http://mic.sgmjournals.org 1811 purL mutant of S. fredii Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 Cosmids pMUS509, pCOS110 and pJN382A were transferred to S. fredii SVQ295. Cosmid pCOS110 carries the purYQL region of R. etli and pJN382A is a derivative of pCOS110 carrying Tn5in purL. SVQ295 transconjugants carrying pMUS509 or pCOS110 were able to grow on MMB, demonstrating that the R. etli purL gene can complement the purL mutation of S. fredii HH103-1. SVQ295 carrying pMUS509, pCOS110 or pJN382A was tested for nodulation ability on soybean cv. Williams. Plants were grown in the presence or absence of thiamin. Soybean plants inoculated with SVQ295(pCOS110) or SVQ295(pMUS509) induced nitrogen-fixing nodules, while those inoculated with SVQ295(pJN382A) only formed Fix 2 pseudonodules. These results demonstrated that the mutation on the purL gene is responsible for the symbiotic-defective phenotype of mutant SVQ295. The presence of thiamin in the plant nutrient solution did not have any significant effect on bacterial nodulation. A3?2kb BamHI fragment of cosmid pMUS509, which positively hybridized to a probe containing 0?5 kb of the SVQ295 DNA that is adjacent to the transposon IS50Rin plasmid pMUS375, was subcloned into pBluescript. From the resultant plasmid a KpnI–BamHI 2870 bp fragment was sequenced (accession number AF275718). By matching this sequence with that obtained from the partial sequencing of pMUS375 using the IS50R primer pdad2, it was possible to position the transposon at nucleotide 832 of the final 2870 bp sequence. The sequenced fragment contains one ORF with a high probability of encoding protein, as indicated by the Testcode algorithm. This ORF encodes a protein that is homologous to several PurL proteins of different organisms such as S. meliloti,Mesorhizobium loti and Agrobacterium tumefaciens (Table 2). purL begins at position 461 and extends for 2129 bp, encoding a deduced polypeptide of Fig. 3. Nodulation phenotype of SVQ295. (A) Morphology of nodules induced by SVQ295 on Glycyrrhiza uralensis. Bar, 1 cm. (B–E) Light microscopy of sections of nodules induced on G. uralensis by HH103-1 (B) and SVQ295 (C), and on Macroptilium atropurpureum by HH103-1 (D) and SVQ295 (E). Magnification: 636in (B–D); 166in (E). 1812 Microbiology 149 A. M. Buendı´a-Claverı´a and others Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 743 amino acids with a predicted size of 79?1 kDa. The ORF is preceded by a putative ribosome-binding site, GAAGGA, at positions 219/214. The sequence AGCGGCCGAACCGGCT at positions 2229/2214 of the putative ATG start codon of the S. fredii purL gene shows similarities to the PurBox reported in Lactococcus lactis (Kilstrup et al., 1998). In addition, the sequence CGCAAAGCGCCCCCTGTTTGTGTTGCGGTTC at positions 290/259 from the ATG codon resembles the E. coli PurBox (GGCAAACGGTTTCGTC) as defined by Sampei & Mizobuchi (1989). DNA sequences upstream and downstream of the identified ORF do not show clear homology to any known gene. The abundance of stop codons in these two regions indicates that they should not contain any coding region. Serological studies on the LPS produced by S. fredii SVQ295 PAGE showed that SVQ295 carrying cosmid pMUS509 has an LPS profile that is indistinguishable from that of HH103-1 (Fig. 4A, lanes 3 and 1 respectively). Four mAbs raised against free-living cultures of HH103 (NB7-242.33) or nodule-derived bacteroids (NB3-4.72; NB3-4.79 and NB6-228.22) and that recognize the same pattern of bands that are visible in silver-stained LPS gels of S. fredii HH103 were tested for their reactivity against the LPS of mutant SVQ295. Fig. 4 shows that the four mAbs reacted with the HH103-1 LPS bands corresponding to the LPS I and LPS II regions (lane 1 of panels B–E) but failed to recognize any LPS band in mutant strain SVQ295 (lane 2 of panels B–E). SVQ295 carrying pMUS509 or pCOS110 produced LPS that were fully recognized by all the assayed mAbs (lanes 3 and 4 of panels B–E). Co-inoculation experiments We investigated the effect of co-inoculating soybean plants with SVQ295 and the S. fredii HH103 nodA mutant SVQ116, which cannot produce nodulation factors (LCOs) (Buendı ´a-Claverı ´aet al., 1996). Hence, the inoculum mixture was composed of a mutant (SVQ295) that produces LCOs but is unable to infect soybean roots and another one (SVQ116) that should be infective but is totally unable to trigger nodule development. Soybean plants growing in a plant nutritive solution containing adenine were also inoculated with SVQ295 or with SVQ295 and SVQ116. As expected, single inoculation with SVQ295 only induced pseudonodules on soybean. In this experiment, however, SVQ295 was able to induce non-nitrogen-fixing nodules on soybean when adenine was added to the plant nutritive solution. Soybean plants inoculated with both mutants (in the presence or absence of adenine) formed a mixture of pseudonodules and true nitrogen-fixing nodules (Table 3). No colonies were formed when nodules were directly crushed on TY medium containing streptomycin (chromosomal marker of SVQ295). All bacterial isolates (900 colonies tested) from nodules crushed on TY medium showed the markers of SVQ116 (Rif r ,Km r , Str s and prototrophy). These results demonstrate that, regardless of the presence or absence of adenine, only the prototrophic LCO 2 mutant SVQ116 was able to penetrate into the plant cells. DISCUSSION Previous reports have clearly shown that purine auxotrophic mutants of (Sino)Rhizobium are symbiotically defective, Table 2. Comparison of HH103 PurL protein with PurL proteins from different organisms. Organism Identity (%) Similarity (%) Amino acids overlapping Length of PurL protein Accession no. Sinorhizobium meliloti 92 96 743 743 Q92PH7 Mesorhizobium loti 81 90 739 743 Q98NN7 Agrobacterium tumefaciens 86 91 769 775 AAL42846 Zymomonas mobilis 61 73 745 734 AAF23789* Bacillus subtilis 48 61 725 742 AAA22679* Synechocystis sp. 48 61 741 777 BAA16646 Lactobacillus lactis 46 61 731 739 AAD12626 Streptomyces coelicolor 42 59 729 752 CAB56359 Mycobacterium tuberculosis 42 56 748 754 P54876 Campylobacter jejuni 44 62 701 728 CAB73212 Lactobacillus casei 43 58 744 741 P35852 Methanococcus jannaschii 41 56 737 733 Q58660 Escherichia coli 25 41 771 1295 P15254 Salmonella typhimurium 24 40 770 1295 AAB08888 Vibrio cholerae 25 41 667 1297 AAF94031 Pseudomonas aeruginosa 24 40 685 1298 AAG07150 *Defined in the protein library as proteins encoded by the purQ gene. http://mic.sgmjournals.org 1813 purL mutant of S. fredii Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 forming Fix 2 pseudonodules that show centrally located vascular bundles (Pankhurst & Schwinghamer 1974; Denarie ´et al., 1976; Pain, 1979; Noel et al., 1984, 1988; Newman et al., 1992, 1995; Djordjevic et al., 1996). A possible explanation for the linking of purine auxotrophy and symbiotic deficiency is that, during nodule formation, adenine might act as a precursor for cytokinins, which are N-6 substituted derivatives of adenine (Pankhurst & Schwinghamer, 1974; Pain, 1979). This possibility does not exclude that adenine auxotrophs might suffer other alterations that, for any reason, could prevent normal nodule development. In fact, purine auxotrophs usually exhibit pleiotropic phenotypes that indicate the occurrence of multiple alterations of the bacterial physiology. For instance, R. etli purine auxotrophs expressed, in free-living Table 3. Co-inoculation of soybean plants cv. Williams with S. fredii mutants SVQ116 and SVQ295 in the presence or absence of adenine Inoculant(s) Nodulation* Appearance of plantsDMarkers of the nodule isolatesd SVQ116 Nod 2 Yellow – SVQ295 Pseudonodules Yellow – SVQ295 +ade§ Pseudonodules and small nodules (8) Yellow Str r Km r , Ade 2 Thi 2 (28) SVQ116 +SVQ295 Pseudonodules and nodules (41) Green Rif r Km r Str s , Ade + Thi + (50) SVQ116 +SVQ295 +ade§ Pseudonodules and nodules (43) Green Rif r Km r Str s , Ade + Thi + (40) HH103-1 Fix + (160) Green Str r Km s , Ade + Thi + (4) *Six plants were used for each treatment. The numbers in parentheses indicate the mean number of nodules per soybean plant. Plants were analysed 82 days after inoculation. DNitrogenase activity of green soybean plants was confirmed by acetylene reduction assay. dThe numbers in parentheses indicate the number of nodules analysed. From each nodule 10 colonies were checked for antibiotic resistance and auxotrophy markers. §Fa ˚hraeus nutritive solution was supplemented with 0?8mg adenine ml 21 . Fig. 4. Silver staining (A) and immunostaining (B–E) of LPS profiles from S. fredii strains using the mAbs NB3-4.72 (B), NB3-4.79 (C), NB-6-228.22 (D) and NB7-242.33 (E). Lanes: 1, HH103-1; 2, SVQ295; 3, SVQ295(pMUS509); 4, SVQ295(pCOS110). 1814 Microbiology 149 A. M. Buendı´a-Claverı´a and others Downloaded from www.microbiologyresearch.org by IP: 150.214.182.233 On: Mon, 14 Mar 2016 13:24:22 conditions, higher levels of the fixNOQP operon, which codes for the symbiotic cytochrome terminal oxidase ccb 3 (Soberon et al., 1997). Moreover, Rhizobium sp. NGR234 purY and purF mutants exhibited multiple differences in gene expression when compared to the parental wild-type strain (Guerreiro et al., 1998). In this investigation we have studied an auxotrophic mutant of S. fredii HH103 that was isolated in a screen for the appearance of alterations in colony morphology. Mutant SVQ295 requires adenine and thiamin for growth in minimal media, which is in agreement with the fact that the reaction catalysed by the PurL enzyme occurs before AICAR in the biosynthetic pathway for purines (Neuhard & Nygaard, 1987). In addition, SVQ295 can also grow in minimal media if the supplementation with adenine is replaced by the purine precursor AICA-riboside. Although SVQ295 cultures grown in the presence of 0?5 mM AICAriboside showed a longer doubling time than that observed in MMB medium with adenine, the final optical density was the same in both bacterial cultures (with AICA-riboside or adenine). Fully grown SVQ295 cultures in the presence of AICA-riboside did not contain ade + revertants, since bacterial subcultures failed to grow in MMB medium devoid of adenine or AICA-riboside. Hence, we conclude that AICA-riboside can indeed act as a purine source for pur mutants. Previous reports showed that pur mutants of S. fredii HH303 and R. leguminosarum bv. viciae 128C53 grew poorly 18 h after the addition of AICA-riboside (Newman et al., 1994). Our results also support this observation, since an initial population of approximately 4610 5 bacteria ml 21 in MMB supplemented with AICAriboside did not show any increase after 5 h incubation and only increased to 7610 5 bacteria ml 21 after 18 h. Hence, the latency period exhibited by SVQ295 cultures in the presence of AICA-riboside and their long doubling time may explain the poor growth of these pur mutants during their first hours of incubation in the presence of AICAriboside. However, after 18 h incubation, a purF mutant of R. etli did not show poor growth in minimal medium supplemented with AICA-riboside (Newman et al., 1994). Mutant SVQ295 failed to nodulate six different legumes (see Results), and formed pseudonodules on G. max.On Macroptilium atropurpureum, small ineffective nodules were formed. Bacterial release in M. atropurpureum nodules was only observed in a few clustered plant cells (Fig. 3E). In Glycyrrhiza uralensis, however, bacterial release was much more extended through the nodule tissue and nitrogenase activity was detected (Fig. 3C). This variable symbiotic phenotype of SVQ295 might be due to the differential ability of each legume host in supplying purine, or purine precursors, to the bacteria during the nodulation process. The fact that G. uralensis forms indeterminate nodules (Fig. 3A) is not relevant to the ability of SVQ295 to infect nodule cells, since Albizia lebbek also forms indeterminate nodules (with the parental strain HH103) but fails to nodulate with the purL mutant. Previous reports have clearly shown that the addition of AICA-riboside to the plant nutritive solution promotes nodulation ability of different rhizobial pur mutants (Newman et al., 1994, 1995; Djordjevic et al., 1996). In addition to supporting these observations our results also show that, in long-term experiments (82 days), the addition of adenine can also promote the formation of small soybean nodules by mutant SVQ295. This late nodulation appears to be the consequence of the exogenous addition of adenine to the plant root medium because, in the absence of adenine, soybean plants inoculated with SVQ295 only formed pseudonodules (Table 3). Mutant SVQ295 carries a transposon Tn5-Mob insertion between nucleotides 831 and 832 of the sequenced DNA fragment, which corresponds to the 59end of a 2690 bp ORF showing homology to the purL gene of S. meliloti. This ORF would encode a putative polypeptide of 743 amino acids, clearly shorter than the E. coli PurL protein (1295 amino acids). The assumption that SVQ295 is mutated in its purL gene is supported by the fact that the mutant was complemented by cosmid pCOS110 (it carries R. etli purYQL genes), but not by cosmid pJN382A (a derivative of pCOS110 carrying a Tn5insertion in the purL gene). In E. coli, the purL gene encodes a 59-phosphoribosylformylglycinamidine synthase (FGAR amidotransferase) that catalyses the conversion of 59-phosphoribosyl-Nformylglycinamide into 59-phosphoribosyl-N-formylglycinamidine (Sampei & Mizobuchi, 1989). In Bacillus subtilis and Lactobacillus casei, however, this enzyme is formed by the products of two different genes, purQ and purL (Ebbole & Zalkin, 1989; Peltonen & Ma ¨ntsa ¨la ¨, 1999). The PurL enzyme in E. coli has three different domains, I, II and III, which have been assigned as potential ATPase, triosephosphate-isomerase-like isomerase, and glutaminase, respectively (Sampei & Mizobuchi, 1989). The E. coli PurL domain III aligns with the B. subtilis and L. casei PurQ protein, while domains I and II align with the PurL protein of B. subtilis,L. casei and S. fredii HH103. This indicates that S. fredii HH103 might also need at least two different genes for the formation of a functional FGAR amidotransferase. Other bacteria, such as Methanococcus jannaschii, Mycobacterium tuberculosis and Synechocystis sp. also appear to have a multi-subunit form of the FGAR amidotransferase enzyme (Bult et al., 1996; Jackson et al., 1996; Kaneko et al., 1996). In all these cases, in which the enzyme appears to be composed of different subunits, the ORF defined as purL encodes a polypeptide of 705–777 amino acids, which is clearly shorter than the FGAR amidotransferase (1295 amino acids) encoded by the purL gene of E. coli (Table 2). The L. casei and B. subtilis purQ and purL genes belong to an operon composed of a cluster of other purine genes, while the E. coli purL gene constitutes a single transcriptional unit which is independent of other purine genes. The purL of S. fredii HH103 also appears to constitute a single transcriptional unit, since putative ORFs were not found http://mic.sgmjournals.org 1815 purL mutant of S. fredii