INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Bacteria of the genera Azorhizobium, Bradyrhizobium, Mesorhizobium, Rhizobium, and Sinorhizobium (commonly called rhizobia) form symbiotic associations with leguminous plants. Establishment of successful symbioses requires signal exchange between the two symbionts. Roots release phenolic compounds that stimulate rhizobia to produce and secrete a class of lipooligosaccharides (LCOs) called Nod factors. In turn, Nod factors induce root hair deformation and the formation of nodule meristems (for reviews, see references 6, 7, and 15). Nod factors are oligomers of three to five (1-4)-N-acetylglucosamine residues, which carry an acyl group at the nonreducing terminus as well as various other substituents (e.g., carbamoyl, O-acetyl, sulfate, or N- or O-methyl groups, etc).

Much of the remarkable ability of Rhizobium species strain NGR234 to nodulate more than 110 genera of legumes, as well as the nonlegume Parasponia andersonii (30, 43), stems from the more than 80 different Nod factors that it secretes. LCOs of NGR234 are pentamers carrying a variety of substituents: the terminal nonreducing glucosamine is N acylated with palmitic, palmitoleic, stearic, or vaccenic acid, is N methylated, and carries one or two carbamoyl groups. The reducing N-acetylglucosamine (GlcNAc) residue is substituted on position 6 with 2-O-methyl-L-fucose, which may be acetylated, sulfated, or nonsubstituted (28). Most of the genes responsible for Nod factor synthesis (nod, nol, and noe) and nitrogen fixation (nif and fix) are carried on the 536-kb symbiotic plasmid pNGR234a (4, 25). Of the 416 predicted open reading frames present, only 20 are directly involved in synthesis of Nod factors (12). Among these are nodABC, which are responsible for the formation of the Nod factor skeleton (35), and various host-specific loci that carry genes responsible for the adjunction of different groups to the core molecules. NodS, for example, is involved in N methylation and NodU and NolO are involved in 6-O and 3-O carbamoylation, respectively (13, 18, 19), nodZ encodes a fucosyltransferase (32), and NoeE is a fucose-specific sulfotransferase (14, 33).

Rhizobial Nod factors may be O acetylated at three distinct sites. In Rhizobium leguminosarum bv. trifolii and R. leguminosarum bv. viciae (40, 41), Rhizobium meliloti 2011 (22), and Bradyrhizobium japonicum USDA135 and Bradyrhizobium elkanii USDA61 (5), C-6 of the nonreducing glucosamine is O acetylated. In R. leguminosarum bv. viciae, a 6-O-acetyltransferase is encoded by nodL (3). 6-O acetylation of the reducing terminus depends on NodX of R. leguminosarum bv. viciae strain TOM, however (11). Perhaps because of the preferences for different Nod factor termini, similarities between NodL and NodX were not found. On the other hand, the fucose residues of Nod factors of Rhizobium etli, Rhizobium loti, and NGR234 are often acetylated (24, 26, 28). In R. loti, the protein encoded by nolL has significant similarity with an O-acetyltransferase of Xanthomonas campestris and other bacteria (38). Since a homologue of nolL (y4eH) is present on pNGR234a, we explored the possibility that this gene is responsible for acetylation of NGR234 Nod factors (NodNGR factors).

Here we show that the expression of y4eH is modulated by flavonoids, and controlled by NodD1, via a functional nod box. Although insertional mutagenesis of y4eH apparently had no effect on nodulation of the plants tested, introduction of nolL into R. fredii USDA257 extended its host range to include Calopogonium caeruleum, Leucaena leucocephala, and Lotus halophilus. Finally, in vitro analyses suggest that y4eH may encode a functional acetyltransferase, NolL, which is responsible for the O acetylation of the fucose residues of NodNGR factors.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Microbiological and molecular techniques. The bacterial strains and plasmids used in this study are listed in Table 1. Strains of Escherichia coli were grown at 37°C in or on Luria-Bertani medium (37), and Rhizobium strains were grown at 27°C in or on Rhizobium minimal medium with succinate as the carbon source (RMS) (4). Antibiotics were used at the following concentrations: ampicillin and rifampin, 100 µg · ml¹; kanamycin and spectinomycin, 50 µg · ml¹; chloramphenicol and gentamicin, 25 µg · ml¹; and tetracycline, 15 µg · ml¹. Purification of plasmid and genomic DNAs, digestion with restriction endonucleases, transformation of bacteria, cloning, and Southern transfers were performed as described by Sambrook et al. (37). Promoter activity of the nod box of y4eH was verified by -galactosidase assays. An 852-bp PstI fragment containing the nolL nod box was cloned into pBluescript-II KS(+), and the insert was then excised with BamHI and KpnI and cloned into the reporter vector pMP220 (39) to generate pNBnolL. This was conjugated into NGR234, NGRnodD1, and NGRsyrM1 by triparental matings. -Galactosidase activity was determined as described by Fellay et al. (9).

Construction of NGRnolL, USDA257(pnolL), and ANU265(pA28, pnolL). Cosmid pXBS23 contains a 2,736-bp XhoI fragment which includes nolL and its promoter (25). After excision with XhoI, this fragment was cloned into the SalI site of a modified pBluescript-II KS(+) vector (Stratagene, La Jolla, Calif.) with the BamHI site deleted. The insert of the resulting clone, pBS-nolL, was excised with SpeI-XhoI, purified on agarose gels, and cloned into the suicide vector pJQ200SK (31). The Km^r  interposon (8) was then inserted into the internal BamHI site of nolL to produce clone pJQ200nolL, which was mobilized into NGR234 by triparental matings with the helper plasmid pRK2013 (10). Marker exchange was selected on RMS plates containing 5% (wt/vol) sucrose (31). Integration of the  interposon at the correct position in NGRnolL was verified by Southern transfer analysis. nolL was liberated from pBS-nolL by digestion with XhoI and cloned into the same site of the broad-host-range vector pBBR1MCS-1 (20) to give pnolL. The transconjugants USDA257(pnolL) and ANU265(pA28, pnolL) were produced by mobilizing pnolL into the corresponding strains by using the helper plasmid pRK2013.

Purification and analysis of Nod factors. Rhizobium strains were grown to an optical density at 600 nm of 1 in RMM3 medium with or without 10⁶ M apigenin (28). Cells were removed by centrifugation at 4°C (6,000 × g for 30 min). Extracellular Nod factors were extracted from the supernatant by reverse-phase chromatography on a C₁₈ column as described previously (18, 28). All lipophilic material retained was eluted with methanol. Nod factors were labeled in vivo by using D-[¹⁴C]glucosamine (53 mCi/mmol; Amersham Pharmacia Biotech, Uppsala, Sweden) in the growth medium (14). Deacetylation of Nod factors was performed as described by Firmin et al. (11). To do this, acetylated NodNGR factors were warmed at 37°C for 1 h in 0.2 M NaOH. The reaction mixture was then neutralized with 0.2 M HCl, loaded onto a C₁₈ reverse-phase Sep-Pak column, washed with H₂O and eluted with methanol. Mass spectra were recorded on an Autospec instrument (Fisons, VG-Analytical, Manchester, United Kingdom) (18). Methylated alditol-acetate derivatives of the LCOs were prepared as described by York et al. (44). After separation on a 15-cm Supelco SP fused-silica column (Hewlett-Packard, Palo Alto, Calif.), the methylated alditol-acetate derivatives were analyzed by gas chromatography-mass spectrometry (GC-MS) on a machine fitted with an electron impact ion source. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker 600-MHz spectrometer (9.4 T) with deuterated dimethyl sulfoxide as the solvent and the secondary reference (¹³C;  = 39.5 ppm).

In vitro acetylation assays. When 100-ml cultures of ANU265 reached an optical density at 600 nm of 0.6, apigenin was added to a final concentration of 10⁶ M. Induction proceeded for 4 h, after which the cells were harvested by centrifugation, washed with 50 mM phosphate (pH 7.2), and resuspended in 5 or 1 ml of the same buffer. Crude extracts were obtained by passing 5-ml portions of the washed cells through a French press two times. Any remaining unbroken cells were removed by centrifugation. Three consecutive cycles of freezing and thawing in liquid nitrogen were used on 1-ml portions of the harvested cells to render them permeable. Acetylation reactions were carried out at 28°C for 4 h in a final volume of either 1 ml (with 200 µl of crude extract) or 10 ml (with 1 ml of permeabilized cells). In addition to the cell suspensions, the reaction mixture contained 50 mM phosphate buffer (pH 7.2), 50 µg of substrate, and 10 µCi of 1-¹⁴C-labeled acetyl coenzyme A (acetyl-CoA) (specific activity, 60 mCi/mmol; DuPont NEN, Boston, Mass.). Afterwards, the reaction products were extracted at 27°C for 12 h with 10 ml (for crude extracts) or 1 ml (for permeabilized cells) of 1:2:2:3 (vol/vol) chloroform-propanol-methanol-water. The supernatant was recovered following centrifugation. After evaporative drying, the residue was dissolved in 5 ml of water and applied to a C₁₈ Sep-Pak cartridge (Waters Corp, Milford, Mass.). The acylated molecules were recovered by elution with methanol, concentrated under vacuum, and separated by reverse-phase thin-layer chromatography (RP-TLC). When fucosylated oligochitins were used as substrates (33), any unreacted acetyl-CoA was removed by passage though a short column containing 2 ml of DEAE-Sephadex A25 (Amersham Pharmacia Biotech) equilibrated in water. Nonionic fractions were concentrated by evaporation under vacuum and spotted on Silica Gel 60 RP-TLC plates (Merck, Darmstadt, Germany). Following development with butanol-ethanol-H₂O (45:30:25), the dried plates were exposed to monitor incorporation of [¹⁴C]acetate into fucosylated oligochitins.

Plant assays. Nodulation capacities of wild-type NGR234, the NGRnolL mutant and the transconjugant USDA257(pnolL) were tested on Calopogonium caeruleum, Lablab purpureus, Leucaena diversifolia, Leucaena leucocephala, Lotus halophilus, Lotus pedunculatus, Pachyrhisus tuberosus, Tephrosia rosea, and Vigna unguiculata as described previously (30). Plants were grown in Magenta jars and harvested at 6 to 8 weeks after inoculation (23).

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Expression of nolL is controlled by a functional NodD1-dependent nod box. The predicted product of open reading frame y4eH (nucleotide positions 96093 to 97193 [12]) shows striking homology to NolL of R. loti (61% identity and 89% similarity). For this reason, y4eH was also named nolL. In NGR234, nolL encodes a 366-amino-acid product with a nonmodified molecular mass of 40.5 kDa and a predicted pI of 9.2 (Swiss-Prot accession no. P55431). nolL is preceded by a well-conserved nod box-like sequence, the promoter activity of which was assayed by cloning an 852-bp PstI fragment of pXBS23 containing the nolL nod box into pMP220 (generating pNBnolL). Only basal levels of -galactosidase activity (ca. 300 Miller units) were detected in liquid cultures of NGR234(pNBnolL), whereas more than 3,400 Miller units were measured at 8 h following induction with apigenin.

Localization of the acetate group on NodNGR factors. Initially, NodNGR factors produced by the overproducing strain NGR(pA28) were used for characterization (28). Later, it became clear that unmodified, wild-type NGR234 produces sufficient amounts of LCOs for chemical analyses (36). As a first step towards fully characterizing acetylation of NodNGR factors, methyl-ester derivatives of the fatty acid chains released from crude C₁₈ extracts of wild-type NGR234 supernatants were analyzed by GC-MS (data not shown). These analyses revealed the following acyl components (relative peak heights are given in parentheses): C_16:1 (0.16), C_16:0 (0.54), C_18:1 (1), C_18:2 (0.12), and OH C_18:1 (0.3). Fast atom bombardment (FAB) MS analysis of the reverse-phase high-pressure liquid chromatography-purified fraction that eluted at 24.5 min produced molecular and fragment ion spectra corresponding to a mixture of LCOs which were N acylated by either C_16:1, C_18:0, C_18:1, C_18:2, or hydroxylated C_18:1, carrying acetylated or nonsubstituted methylfucose. Ions corresponding to acetylated, C_16:0-acylated molecules were not observed. GC-MS of the alditol-acetate derivatives revealed the presence of the monosaccharides methylfucose, GlcNAC, and N-methylglucosamine, as well as traces of fucose, N-acetylmannosamine, and N-acetylgalactosamine (Fig. 1). The structures and relative abundances of acetylated NodNGR factors are reported in Table 2.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   GC profiles of alditol-acetate derivatives of LCOs extracted from supernatants of NGR234 cultures. Carbohydrate constituents are methylfucose (MeFuc), fucose (Fuc), N-acetylglucosamine (GlcNAc), N-acetylmannosamine (ManNAc), N-acetylgalactosamine (GalNAc), and N-methylglucosamine (NMeGlcN).

View larger version (28K): [in this window] [in a new window]   FIG. 2.   One- and two-dimensional COSY 1H NMR spectra of the major acetylated NodNGR factors. The signals at  = 2.02 and 2.06 ppm correspond to acetate groups at O-4 or O-3 of the fucose which provoke splitting of the fucose proton signals (A and B correspond to the positions of the protons when they are in O-4 or O-3, respectively). Coupling between H-6 and H-5 is seen as a cross-peak when the acetate is at O-3 or O-4 (A6-A5 and B6-B5), but coupling between H-5 and H-4 is not seen because H-5-C-C-H-4 should form a right angle. Similar behavior between B3 and B4 is observed only when the acetate is on O-3. When the acetate group is on O-4, coupling is seen by a cross-peak between A4 and A3.

Nod factors produced by the NGRnolL mutant. To test whether acetylation of NodNGR factors depends on a functional nolL gene, an  interposon was inserted into this gene. Nod factors produced by the mutant (NGRnolL) were purified by high-pressure liquid chromatography and characterized by FAB MS. In the negative-ionization mode, no significant differences were observed in the sulfated products collected after 17.5 min of elution, compared to published spectra (28). In the positive-ionization mode, wild-type NodNGR factors collected at 24.5 min correspond to a mixture of molecules in which the methylfucose group was either acetylated or nonacetylated (28). In contrast, when purified from cultures of NGRnolL, this 24.5 min fraction yielded molecular ions [M + H]⁺ which were 42 Da lighter and correspond only to nonacetylated products. Furthermore, ¹H NMR analysis of this same peak produced a spectrum similar to that observed after mild alkaline hydrolysis (Fig. 3A), showing the loss of signals at  = 2.06 and 2.02 ppm which are characteristic of an acetate group at position O-3 or O-4 of the fucose.

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Selected regions of the 1H NMR spectra showing the signals originating from the acetyl residue (Ac). (A) Nonacetylated Nod factors from NGRnolL. R. fredii Nod factors and chemically deacetylated LCOs have the same 1H NMR spectra in this region. (B) Acetylated NodNGR factors. (C) LCOs of USDA257(pnolL). (D) LCOs of NGRnoeI, in which noeI, the gene required for 2-O-methyltransferase activity, is disrupted (19). This mutant produces nonmethylated but partly acetylated LCOs. These spectra show that the acetate group on the fucose migrates from a hydroxyl to a vicinal hydroxyl residue, as shown by the number of acetate signals. When free, the 2-O position of fucose can also be occupied by acetate, giving a third signal. In NodNGR factors, R. fredii Nod factors, and LCOs produced by NGRnolL, the 2-O position of fucose is mostly (95%), partially (70%), or not methylated, respectively.

Nod factors produced by USDA257 containing the nolL gene of NGR234. The 2.7-kb XhoI fragment of pXBS23 containing nolL together with its nod box promoter was cloned into the broad-host-range vector pBBR1MCS-1, generating pnolL (Table 1). Introduction of pnolL into USDA257, purification of the Nod factors, and their analysis by FAB MS revealed low-mass fragments at m/z 428, 426, and 328, which are similar to those observed with LCOs produced by wild-type USDA257 (2, 18). This suggests that the nonreducing termini of both types of Nod factors were not modified. In contrast, pseudo-molecular ions were shifted up by 42 Da, which corresponds to the presence of an additional acetate group on the fucose and methylfucose (Fig. 4). Acetylation of C_16:1, C_18:0, and C_18:1 N-acylated penta-, tetra-, and trimeric Nod factors was observed (data not shown). Ion fragments from the pseudo-molecular ions (m/z 1458 to 1416) which lost 42 mass units confirmed the position of the additional acetate group on the fucose residue (Fig. 4). Minor traces of nonacetylated Nod factors, like those secreted by wild-type USDA257, were also found.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   FAB MS of the major Nod factors produced by USDA257(pnolL) transconjugants. The molecular ion [M + H]+ at m/z 1458.7 and its adduct [M + Na]+ at m/z 1480.7 correspond to pentameric LCOs substituted with vaccenic acid (C18:1) and acetylmethylfucose or acetylfucose (m/z 1444.8).

In vitro activity of NolL. In vitro enzyme assays were performed to confirm whether NolL is essential for transacetylation activity. To do this, pnolL was introduced into ANU265(pA28) (NGR234 cured of pNGR234a but containing nodD1 on a low-copy-number plasmid), generating ANU265(pA28, pnolL). Crude extracts of apigenin-induced cultures of the transconjugant as well as the control strain ANU265(pA28) were prepared by passage through a French press. Following incubation with ¹⁴C-labeled acetyl-CoA (the acetate donor) and NodNGR factors purified from the NGRnolL mutant (the candidate acetyl acceptors), the reaction products were separated by RP-TLC analysis of the lipophilic extracts. Unfortunately, acetyltransferase activity was not detected under these conditions (data not shown). Acetyltransferase activities of crude extracts and permeabilized cells were also assayed by using fucosylated oligochitins [(GlcNAc)₁Fuc to (GlcNAc)₆Fuc] (33) as substrates. Again, radioactivity was not incorporated into fucosylated oligochitins as shown by RP-TLC analyses.

View larger version (38K): [in this window] [in a new window]   FIG. 5.   C18 RP-TLC analysis of acetyltransferase activity in vitro. Nonacetylated NodNGR factors were incubated with 14C-labeled acetyl-CoA along with induced permeabilized cells of either ANU265(pA28, pnolL) (lane 2) or ANU265(pA28) (lane 3). The standard (lane 1) represents NodNGR factors labeled with [14C]glucosamine in vivo by whole NGR234 cells (spot A, acetylated or not substituted; spot B, sulfated). Plates were developed with methanol-ammonia (9:1) and exposed to X-ray film. The spots above spot B result from unincorporated [14C]glucosamine.

nolL of NGR234 behaves as a host specificity gene in USDA257. Variously substituted Nod factors are determinants of host specificity. R. fredii USDA257 excretes an exact subset of the NodNGR factors and nodulates an exact subset of the NGR234 hosts (30). In this context, it is surprising that inactivation of nolL in NGR234 did not modify the host range of the mutant strain, but conjugation of nolL into USDA257 broadened its spectrum of hosts to include C. caeruleum, Leucaena leucocephala and Lotus halophilus (Table 3). The reasons for these varying phenotypes are not clear, especially since USDA257(pnolL) had no effect on P. tuberosus and T. rosea.

View this table: [in this window] [in a new window]   TABLE 3.   Effect of Rhizobium sp. strain NGR234, NGRnolL, R. fredii USDA257, and USDA257(pnolL) transconjugant on nodulation of a number of legumes possessing determinate (Calopogonium, Lotus, Pachyrhizus, and Vigna) and indeterminate (Leucaena and Tephrosia) nodules

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Most fucose residues of NodNGR factors are 2-O-methylated and usually acetylated but are only partially sulfated. Physical and chemical analyses of LCOs secreted by both the NGRnolL mutant and the USDA257(pnolL) transconjugant confirmed that acetylation is dependent on a functional y4eH locus. Unlike nodSU, which is present but inactive in USDA257 (21), no gene homologous to y4eH was identified in this strain (data not shown). Mobilization of nolL into USDA257 led not only to acetylation of the fucose residue but also to an extended host range. On the other hand, inactivation of nolL did not appear to reduce the nodulation capacity of NGR234 mutant (only a limited subset of the NGR234 hosts was tested).

In contrast to the case for NGR234, mutation of nolL of R. loti causes the loss of nodulation of Leucaena leucocephala and Lotus pedunculatus (38). Of course, Nod factors of the two rhizobia are not identical. The numbers and locations of the carbamoyl groups are different, and the fucose is not methylated in R. loti Nod factors (24, 28). Perhaps because of this, the host ranges of the two mutants are not identical. Nevertheless, it is difficult to correlate variation in substitutions of the reducing terminus with host range. Undoubtedly, levels of Nod factors are also determinants of host specificity (36).

Experiments in which nolL and its promoter were mobilized into ANU265(pA28) suggest that NolL is required for transacetyltransferase activity. Unfortunately, confirmation that the purified NolL protein is the acetyltransferase has been difficult to obtain. Cell extracts of ANU265(pA28, pnolL), prepared by using a French press, did not retain transacetylation activity. Activity was detected only in permeabilized cells of the transconjugants, suggesting that disruption of the cytoplasmic membranes results in loss of enzyme activity. Indeed, BLAST alignments (1) suggest the presence of nine transmembrane domains in the NolL protein (data not shown). It is thus probably impossible to separate the active enzyme from the membrane fraction.

Since fucosylated oligomers of chitin (di- to pentamers) were unable to accept acetate from CoA, acylated LCOs (i.e., Nod factors) must be the preferred substrates of NolL. Similar results were obtained for NoeE, a sulfotransferase (33), and suggest that substitution of the fucose occurs after biosynthesis of the Nod factor core. Most probably, the presence of acetate groups at positions O-3 and O-4 is not due to the lack of stereospecificity of NolL but rather to the ability of the acetate group to migrate from one hydroxyl group to another free -hydroxyl group. Support for this suggestion comes from the observation that when the C-2 hydroxyl is free in LCOs produced by NGRnoeI (which lack the methyl group [19]), acetate migrates to this position. ¹H NMR of NodNGRnoeI factors shows an additional signal at 2.04 ppm, which is expected of the chemical shift of CH₃COO- at O-2 (Fig. 3D). Surprisingly, the ratio of sulfated (Fig. 5, lane 1, spot B) to nonsulfated (spot A) NodNGR factors is not affected by mutation of nolL (data not shown). Perhaps the proportions of the different Nod factors are limited by the concentrations of substrates rather by competition between the enzymes NoeE and NolL (27).

Little homology exists between the different acetyltransferases of various rhizobia. Perhaps because NodX and NodL acetylate the reducing- and nonreducing termini of R. leguminosarum and R. meliloti Nod factors, respectively, no significant homology exists between their genes. Similarly, the NolL proteins of both R. loti and NGR234 utilize fucosylated Nod factors as substrates. Thus, the primary substrates of all three enzymes are different. Possibly because of this, the amino acid sequences have not been well conserved. Furthermore, NodL is a cytoplasmic protein which belongs to the family of acetyltransferases characterized by the hexapeptide repeat (LIV)-G-X₄. This signature is found neither in NolL nor in NodX, which are intimately associated with membranes.