Biosynthesis of Azorhizobium caulinodans Nod Factors

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1995 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 270, No. 49, Issue of December 8, pp. 29217–29223, 1995 Printed in U.S.A.

Biosynthesis of Azorhizobium caulinodans Nod Factors STUDY OF THE ACTIVITY OF THE NodABCS PROTEINS BY EXPRESSION OF THE GENES IN ESCHERICHIA COLI* (Received for publication, July 6, 1995, and in revised form, September 26, 1995)

Peter Mergaert‡, Wim D’Haeze‡, Danny Geelen‡, Danielle Prome´§, Marc Van Montagu‡¶, Roberto Geremiai, Jean-Claude Prome´§, and Marcelle Holsters‡** From the ‡Laboratorium voor Genetica, Universiteit Gent, B-9000 Gent, Belgium, the §Laboratoire de Pharmacologie et de Toxicologie Fondamentales, CNRS, F-31077 Toulouse Cedex, France, and the iCentre de Recherches sur les Macromolećules Ve´ge´tales, CNRS, F-38041 Grenoble Cedex, France

By in vitro and in vivo studies with Escherichia coli expressing different combinations of the nodABCS genes of Azorhizobium caulinodans, Nod factor intermediates were identified and their structures determined using mass spectrometry. Substrate-product relationships were studied by time course experiments, and the Nod factor biosynthetic pathway was partially resolved. E. coli strains, harboring nodA and/or nodB, did not produce Nod metabolites, whereas the strain expressing nodC produced chitooligosaccharides. Thus, the first committed step was the production of the carbohydrate backbone. Bacitracin and tunicamycin did not affect this step, suggesting that undecaprenyl pyrophosphatelinked intermediates were not involved. The second step was the deacetylation of chitooligosaccharides by NodB since the E. coli strain expressing nodBC produced chitooligosaccharides, deacetylated at the non-reducing end and since the NodC products were precursors of the NodBC products. A strain expressing nodBCS produced N-methylated oligosaccharides, whereas a strain expressing nodCS produced unmethylated oligosaccharides. Time course experiments showed that methylation occurred after deacetylation. Thus, NodS acted after NodB. The NodBCS metabolites were partially converted to lipo-chitooligosaccharides when the nodABCS genes were expressed, showing that NodA was involved in the acylation and acted after NodS.

The unique ability of rhizobia to nodulate leguminous plants largely depends on signal molecules that trigger the noduledevelopmental program. These regulatory molecules, called Nod factors, consist of a lipo-chitooligosaccharide (LCO)1 core, modified with several substitutions (1). The core contains a trito pentamer of b-1,4-linked GlcNAc residues and a fatty acyl * This work was supported in part by grants from the Belgian Program on Interuniversity Poles of Attraction (Prime Minister’s Office, Science Policy Programming, no. 38), the Vlaams Actieprogramma Biotechnologie (ETC 002), and the European Communities’ BIOTECH Program as part of the Project of Technological Priority 1993–1996 (BIO2-CT93-0400). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed: Laboratorium voor Genetica, Universiteit Gent, K. L. Ledeganckstraat 35, B-9000 Gent, Belgium. Tel.: 32-9-2645170; Fax: 32-9-2645349. ** Research Director of the National Fund for Scientific Research (Belgium). 1 The abbreviations used are: LCO, lipo-chitooligosaccharide; CID, collision-induced dissociation; FAB, fast atom bombardment; HPLC, high pressure liquid chromatography; MS, mass spectrometry; AdoMet, S-adenosyl-L-methionine.

moiety that replaces the N-acetyl group of the nonreducing end. This structure is common to all Nod factors, independent of their origin. However, the type of modifications, the length of the sugar chain, and the nature of the fatty acid differ from strain to strain. For example, the Nod factors produced by Azorhizobium caulinodans strain ORS571 are pentamers carrying a vaccenoyl or stearoyl chain. On the nonreducing end, the oligosaccharide is modified with a N-methyl and O-carbamoyl group, and the reducing end is branched with a D-arabinosyl sugar (2). For the production of Nod factors, rhizobia are dependent on their nodulation (nod) genes. The nodABC genes are absolutely required for the synthesis of Nod factors and, hence, for nodulation. NodC produces chitooligosaccharides (3, 4), NodB removes the N-acetyl group of their nonreducing end (5), and NodA adds an acyl chain (6, 7). Modifications of the Nod factor core are dependent on nod genes that are strain specific. A. caulinodans has an operon containing the genes nodABCSUIJZorf9 (8, 9).2 We have shown that the nodS gene is involved in Nod factor methylation and that the NodS protein is an AdoMet-dependent methyl transferase that uses deacetylated chitooligosaccharides as substrate (9, 10). nodU is involved in carbamoylation of Nod factors and nodZ and/or other downstream-located genes in arabinosylation.2 The biochemical activities were ascribed to the NodABCS proteins based on in vitro experiments using synthetic substrates. The occurrence of the substrates and/or products had not been demonstrated in vivo. Here, we describe the isolation and chemical characterization of Nod factor intermediates produced in Escherichia coli expressing nodABCS in different combinations. Our results allow us to propose a biosynthetic pathway for Nod factors in A. caulinodans. EXPERIMENTAL PROCEDURES

Bacterial Strains and Growth Conditions—The A. caulinodans strains ORS571(pRG70), ORS571–1.59S(pRG70 –1.59S), and ORS571– 1.31U(pRG70 –1.31U) (2, 9) were grown in YEB medium at 37 °C. E. coli strains were grown in Luria-Bertani medium without glucose (LB2) at 37 °C (11). Cloning Techniques—The nod genes were cloned in the vector pUC18 or pUC19, downstream from the lacZ promoter (11). Fragments were obtained as restriction fragments or by polymerase chain reaction amplification using the plasmids pRG701 (8) and pSG3–1 (9) as DNA source or template. The primers for polymerase chain reaction were 38, CGGCTGATGATCGACACTGAATTC; 35, GCTGGCGTCACGCTGAATTCCGGC; 39, TTGCGCTTCAGCCGATATCATGAG; 36, TGCGCTCGCCAAGGTGAATTCGGC; and 37, TCGCGCGTCGAGATCCTTGCATGC (Fig. 1). The plasmids (Fig. 1) were introduced into E. coli strain DH5a. 2 D. Geelen, M. Fernańdez-Lo´pez, W. D’Haeze, P. Mergaert, M. Van Montagu, J.-C. Prome´, and M. Holsters, submitted for publication.

29217

29218

Nod Factor Biosynthesis in A. caulinodans

In Vivo Labeling of Nod Metabolites—Strains were grown overnight, diluted to A600 5 0.1 in 1 ml of fresh medium, precultured for 1 h, and induced with 20 mM naringenin for ORS571 derivatives or with 1 mM isopropyl-b-D-thiogalactoside for DH5a derivatives. 2 h after induction, radioactive precursor was added (25 mCi/ml [2-14C]acetate (56 mCi/ mmol), 1 mCi/ml [1-14C]GlcNAc (57 mCi/mmol), or 25 mCi/ml [methyl3 H]methionine (86 Ci/mmol) (radiochemicals were from Amersham, Aylesbury, United Kingdom)). 12 h after induction, the cultures were centrifuged, and the cell pellet was resuspended in 500 ml of water 1 500 ml of n-butyl alcohol. This mixture was vigorously shaken during 1 h. The two phases were separated, and the water phase was extracted a second time. The water phase was loaded on a Sephadex A-25 anion exchange column (Pharmacia Biotech, Uppsala, Sweden) (gel volume, 500 ml) equilibrated with water. The column was washed with 2 ml of water, and the unbound fraction was vacuum dried. Samples were analyzed by silica gel TLC (3). The 14C-labeled standards were obtained by reacetylation of hydrolyzed chitin with [1-14C]acetic anhydride (30 mCi/mmol, Amersham). The n-butyl alcohol phase was washed three times with 100 ml of water, vacuum dried, and dissolved in 500 ml of water. This solution was washed three times with 500 ml of ethyl acetate and vacuum dried. The samples were analyzed by octadecyl silica TLC (2). For labeling experiments in the presence of antibiotics, tunicamycin (10 mg/ml, Sigma) or bacitracin (150 mg/ml, Sigma) were added at the time of induction. For the experiment, described in Fig. 6, a DH5a(pUCNABCS) culture was grown at 25 °C in the presence of [14C]GlcNAc (5 mCi/ml). The incubations were stopped at different time points. Quantification was done with a PhosphorImaging system from Molecular Dynamics, using the Image Quanty software. In Vitro Synthesis of Labeled Nod Metabolites—In vitro incubations were as described (3), except that incubations were at 30 °C for 90 min (except where stated otherwise) and no PPi was added. When AdoMet was included, a final concentration of 1 mg/ml was used. Tunicamycin or bacitracin concentrations were as described above. Reacetylation of deacetylated chitooligosaccharides was done as before (3). For the incubations described in Fig. 5B, UDP-[14C]GlcNAc was omitted, and 14 C-labeled GlcNAc5 ('4 mCi/mmol) was added (20,000 cpm per reaction). Purification of Nod Metabolites—4 liters of cultures, grown in LB2 medium overnight at 37 °C, were centrifuged, and the cells were resuspended in 100 ml of water, boiled during 5 min, and extracted with 100 ml of n-butyl alcohol. For the strains DH5a(pUCNC), DH5a(pUCNBC), and DH5a(pUCNBCS), the water phase was retained. For DH5a(pUCNABCS), the n-butyl alcohol extraction was repeated once, and the combined n-butyl alcohol phases were retained. Chromatographic purifications were followed using radioactively labeled metabolites. The NodABCS metabolites were further purified as was done for Nod factors of ORS571(pRG70) (2). The elution times of the NodABCS metabolites in the final HPLC purification step are given in Table I. The aqueous solutions containing the NodC, NodBC, or NodBCS metabolites were vacuum concentrated to 10 ml, and trichloroacetic acid was added to a final concentration of 10%. The precipitate was centrifuged, and the supernatant was concentrated to '1 ml. This was loaded on a Bio-Gel P-2 column (Bio-Rad; 90 3 1.5 cm, 20 ml/h, eluent 0.1 M pyridine-acetate in water, pH 6.0), and fractions of 2 ml were collected. Nod metabolites were finally purified using an Econosil NH2 5-ml, 250 3 4.6-mm HPLC column (Alltech, Deerfield, IL) with a gradient from 80 to 50% CH3CN in 30 min. Elution times (in minutes) of the different Nod metabolites are 11.8, 15.0, 18.3, and 20.1 for the NodC metabolites GlcNAc2, GlcNAc3, GlcNAc4, and GlcNAc5, respectively, 24.4 for the NodBC metabolite GlcNGlcNAc4, and 27.5 for the NodBCS metabolite GlcNCH3GlcNAc4. Analytical Methods—Mass spectrometry was done with an AutoSpec 6F instrument (VG Analytical, Manchester, UK) equipped with a cesium bombardment source working at 130 kV. The matrix was either a 1:1 mixture of glycerol and 10% trichloroacetic acid for GlcNAc oligomers and their analogues or a 1:1:1 mixture of meta-nitrobenzyl alcohol, glycerol, and 10% trichloroacetic acid for LCOs. The energy of the secondary ion beam was 8 kV. For MS/MS, the collision cell was floating at 4 kV. Helium was the collision gas. Its pressure was adjusted as to reduce the beam to 50% of its original value. Constant ratio B/E scans were performed and averaged. Mass calibration was done using a Hall probe array. Fatty acids were analyzed as methyl esters with gas chromatography using a 5890A instrument (Hewlett-Packard, Wilmington, DE)

FIG. 1. Map of the different constructs used in this study. The nod genes located on the different plasmids are indicated in the name. The constructs were introduced into the E. coli strain DH5a. The numbers indicate the position of the primers used for cloning. Restriction enzymes used are Ba, BamHI; M, MscI; Sa, SalI; P, PstI; Pp, PpuMI; and Sp, SphI. bp, base pairs. equipped with an Ultra 2 silica capillary column coated with methyl phenyl silicone (25 m 3 0.2 mm 3 0.33 mm; Hewlett-Packard). RESULTS

The NodC Metabolites—In our previous study, we showed that the NodC protein synthesizes chitooligosaccharides with a polymerization degree up to five. These results were based on in vitro synthesis by protein extracts of ORS571(pRG70) or DH5a(pUCNC) of products that had the same chromatographic behavior as chitooligosaccharides and that were chitinase degradable (3). However, the in vitro production of NodC metabolites was too low to allow structural analysis. Therefore, in vivo synthesis of NodC metabolites in the strain DH5a(pUCNC) (Fig. 1) was studied. Using [14C]GlcNAc or [14C]acetate, we showed that in vivo produced NodC metabolites were similar to in vitro produced products (Fig. 2, lanes A4 and C2). The in vivo production was scaled up to purify the NodC products. Chromatographic fractions containing putative GlcNAc2, GlcNAc3, GlcNAc4, and GlcNAc5 were analyzed with FAB-MS and CID-MS. The CID-MS spectrum of the GlcNAc5 product, shown in Fig. 3A, displayed a [M1H]1 pseudo-molecular ion at m/z 1034 and fragment ions, formed by sequential glycosidic cleavages at m/z 813, 610, 407, and 204. The same spectrum was obtained with standard GlcNAc5. Also for the fractions that contained the putative GlcNAc4, GlcNAc3, and GlcNAc2, spectra were obtained in accordance to the assumed identity of the products. The ions were m/z 831 ([M1H]1), 610, 407, and 204 for GlcNAc4, m/z 628 ([M1H]1), 407, and 204 for GlcNAc3, and m/z 425 ([M1H]1) and 204 for GlcNAc2. Bacterial polysaccharides such as peptidoglycan and exopolysaccharides are synthesized on a undecaprenyl phosphate lipid carrier (12). We investigated whether the NodC metabolites are also synthesized on such a carrier. Several attempts to isolate lipid-linked intermediates failed. Moreover, the antibiotics tunicamycin, which inhibits the transfer of GlcNAc-1phosphate from UDP-GlcNAc to undecaprenyl phosphate, and bacitracin, which inhibits the dephosphorylation of undecaprenyl pyrophosphate, had no effect on the synthesis of the NodC metabolites neither in vitro nor in vivo (data not shown). The NodBC Metabolites—Under certain conditions such as higher temperatures or longer incubation times, it was observed that in the in vitro incubation reaction using ORS571(pRG70) protein extracts, the NodC metabolites were absent, and new, slower migrating products appeared (Fig. 2C, lane 6). The possibility that these new products were modified forms of the NodC metabolites was investigated with a time course experiment. TLC analysis showed that after 5 min, the


29219

FIG. 2. In vivo and in vitro synthesis of radioactively labeled Nod metabolites. A, direct phase TLC analysis of water-soluble products, produced in vivo from [14C]acetate. Nod metabolites were produced by the following: lane 1, DH5a(pUCNA); lane 2, DH5a(pUCNB); lane 3, DH5a(pUCNAB); lane 4, DH5a(pUCNC); lane 5, DH5a(pUCNBC); lane 6, DH5a(pUCNABC); lane 7, DH5a(pUCNBCS); lane 8, DH5a(pUCNABCS); lane 9, chitooligosaccharide standards. B, reverse-phase TLC analysis of butyl alcohol-soluble products, produced in vivo from [14C]acetate (lanes 1– 8, 13 and 14), [14C]GlcNAc (lanes 9 –10), or [3H]methionine (lanes 11–12). Nod metabolites were produced by the following: lane 1, DH5a(pUCNA); lane 2, DH5a(pUCNB); lane 3, DH5a(pUCNAB); lane 4, DH5a(pUCNC); lane 5, DH5a(pUCNBC); lane 6, DH5a(pUCNBCS); lane 7, DH5a(pUCNABC); lane 8, DH5a(pUCNABCS); lane 9, DH5a(pUCNABC); lane 10, DH5a(pUCNABCS); lane 11, DH5a(pUCNABC); lane 12, DH5a(pUCNABCS); lane 13, ORS571 (induced with 10 mM naringenin); lane 14, ORS571 (uninduced). LCO indicates the position of lipo-chitooligosaccharides. C, direct-phase TLC analysis of in vitro synthesized Nod metabolites from UDP-[14C]GlcNAc. Lane 1, chitooligosaccharide standards; Nod metabolites were produced by protein extracts from the following: lane 2, DH5a(pUCNC); lane 3, DH5a(pUCNBC); lane 4, DH5a(pUCNBCS); lane 5, DH5a(pUCNBCS) in the presence of AdoMet; lane 6, ORS571(pRG70); lane 7, ORS571(pRG70) in the presence of AdoMet; lane 8, ORS571–1.59S(pRG70 –1.59S) in the presence of AdoMet; lane 9, ORS571–1.31U(pRG70 –1.31U) in the presence of AdoMet. Arrow 1 indicates the position of GlcNAc5, arrow 2 of GlcNGlcNAc4, and arrow 3 of GlcNCH3GlcNAc4. Arrowheads indicate the origin of the TLCs.

NodC metabolites could be detected, and they increased to a maximum at 15 min. Thereafter, the amount declined again (Fig. 4). A novel product, migrating just below the NodC metabolites, appeared with increasing intensity between 5 and 45 min (Fig. 4). This pattern was observed for the di-, tri-, tetra-, and pentamers and strongly suggested that after their synthesis, the NodC metabolites are modified to these novel products. The NodB protein of Rhizobium meliloti deacetylates the nonreducing end residue of chitooligosaccharides (5). We investigated whether the novel metabolites were the deacetylated forms of the NodC metabolites and whether they were made by NodB. The in vitro and in vivo synthesis of the NodC metabolites and/or the novel metabolites was analyzed in the strains DH5a(pUCNBC) and DH5a(pUCNB) (Fig. 1). These experiments (Fig. 2, lanes A2, A5, and C3) showed that indeed the NodB protein was involved in the synthesis of the novel metabolites and that their synthesis also depended on the NodC protein. Therefore, they were called the NodBC metabolites. Chemical N-acetylation of the NodBC metabolites resulted in the formation of products that migrated on TLC (Fig. 4B) or HPLC (data not shown) at the position of GlcNAc1,2,3,4,5 standards and NodC metabolites, suggesting that the NodBC metabolites are indeed deacetylated chitooligosaccharides. To prove this and to localize the GlcN residue, the NodBC metabolites were synthesized in vivo by the DH5a(pUCNBC) strain. The putative pentasaccharide was purified and analyzed by mass spectrometry. The [M1H]1 pseudo-molecular ion was determined at m/z 992 in agreement with the mass of a pentasaccharide of four GlcNAc residues and one GlcN residue. The CID-MS spectrum (Fig. 3B) showed fragment ions at m/z 771, 568, 365, and 162. This fragmentation corresponded to GlcNAc5, which is deacetylated at the nonreducing end. The NodBCS Metabolites—We found that the NodS protein

of A. caulinodans and Rhizobium strain NGR234 is an AdoMetdependent methyl transferase involved in Nod factor methylation (9, 10). When AdoMet was included in the in vitro synthesis of Nod metabolites, using ORS571(pRG70) protein extract, a novel product was formed with a lower mobility than the NodBC metabolite (Fig. 2C, lanes 6 and 7). The strain DH5 a (pUCNBCS) (Fig. 1) was compared with DH5a(pUCNBC) for the in vitro and in vivo production of this new metabolite (Fig. 2A, lanes 5 and 7; Fig. 2C, lanes 4 and 5). Also, protein extracts of the nodS mutant ORS571– 1.59S(pRG70 –1.59S) and the nodU mutant ORS571– 1.31U(pRG70 –1.31U) were tested (Fig. 2C, lanes 8 and 9). These results showed indeed that the metabolite was depending on the presence of the nodS gene and was, therefore, the NodBCS metabolite. The chemical structure of a pentameric NodBCS metabolite was determined by FAB-MS and CID-MS (Fig. 3C). The [M1H]1 pseudo-molecular ion was at m/z 1006, and fragment ions were at m/z 785, 582, 379, and 176. These ions were 14 mass units higher than the ions of the NodBC metabolites, in agreement with the presence of a methyl group at the nonreducing end of a deacetylated chitopentasaccharide. The NodS protein methylates in vitro deacetylated chitooligosaccharides but not chitooligosaccharides (10). An in vivo labeling experiment with the strains DH5a(pUCNBCS), DH5a(pUCNCS) (Fig. 1), and DH5a(pUCNBC) and using the precursors [14C]GlcNAc or [3H]methionine (methionine is a precursor for AdoMet) was performed (Fig. 5A). The three strains incorporated the [14C]GlcNAc label in chitooligosaccharides. In contrast, the [3H]methionine label was only incorporated in the chitooligosaccharides produced by DH5a(pUCNBCS). This observation showed that methylation by NodS needs the prior action of NodB. An in vitro time course

29220


FIG. 3. CID mass spectra of Nod metabolites. A, NodC metabolites with [M1H]1 at m/z 1034. B, NodBC metabolites with [M1H]1 at m/z 992. C, NodBCS metabolites with [M1H]1 at m/z 1006. D, NodABCS metabolites with [M1H]1 at m/z 1244.

FIG. 4. Conversion of NodC metabolites to NodBC metabolites. A, TLC analysis of an in vitro time course experiment showing conversion of NodC metabolites to NodBC metabolites in function of time. Protein extracts of ORS571(pRG70) were incubated with UDP-[14C]GlcNAc at 30 °C. Reactions were stopped at different time points indicated in minutes under the lanes. B, TLC analysis of an in vitro incubation with ORS571(pRG70) protein extract at 30 °C during 90 min. The reaction mixture was divided in two parts. One part served as control (2); the second part was chemically N-acetylated (1). Positions of chitooligosaccharide standards are indicated with a bar. The arrowheads indicate the origin of the TLCs.

experiment further confirmed this. As substrate, chemically synthesized [14C]GlcNAc5 was used. TLC analysis showed that the substrate was first converted to the deacetylated form and

then to the methylated, deacetylated form (Fig. 5B). The NodABC and NodABCS Metabolites—The NodA protein of R. meliloti is involved in the in vitro transfer of a fatty acid to deacetylated chitooligosaccharides (6, 7). Our attempts to form LCOs from in vitro synthesized deacetylated chitooligosaccharides using protein extracts from ORS571(pRG70) or DH5a(pUCNA) (Fig. 1) were unsuccessful even when substrates with high specific activity (about 100 Ci/mmol) were used. Therefore, the involvement of the azorhizobial NodA protein in acylation was studied by in vivo labeling experiments using DH5a(pUCNABC) and DH5a(pUCNABCS) (Fig. 1). Radioactive precursors used were [14C]acetate, [14C]GlcNAc, or [3H]methionine, and the production of lipo-chitooligosaccharides was analyzed by reverse-phase TLC. Both strains incorporated [14C]acetate or [14C]GlcNAc into molecules that comigrated on the TLC system with the Nod factors produced by ORS571(pRG70) (Fig. 2B). With [3H]methionine, however, these metabolites could only be labeled in the strain DH5a(pUCNABCS). The DH5a strains containing nodA, nodAB, nodB, nodC, nodBC, or nodBCS did not produce these Nod metabolites (Fig. 2B). These results demonstrated that NodABC was absolutely required and sufficient for the production of the Nod factor core and that the NodA protein was involved in the acyl transfer. The NodABCS metabolites were purified from DH5a(pUCNABCS) cells. The HPLC chromatogram of the final


29221

FIG. 5. Conversion of NodBC metabolites to NodBCS metabolites. A, TLC analysis of Nod metabolites produced in vivo from [14C]GlcNAc (lanes 2– 4) or [3H]methionine (lanes 5–7). Lane 1, chitooligosaccharide standards; Nod metabolites were extracted from the following: DH5 a (pUCNCS) (lanes 2, 5), DH5 a (pUCNBC) (lanes 3, 6), and DH5a(pUCNBCS) (lanes 4, 7). B, TLC analysis of an in vitro time course experiment showing conversion of NodC metabolites to NodBC metabolites to NodBCS metabolites in function of time. Protein extracts of ORS571(pRG70) were incubated with [14C]GlcNAc5 at 30 °C. Reactions were stopped at different time points indicated in minutes under the lanes. In the first lane, chitooligosaccharide standards were loaded. The positions of 1 (GlcNAc5), 2 (GlcNGlcNAc4), and 3 (GlcNCH3GlcNAc4) are indicated with arrows. The arrowheads indicate the origin of the TLCs.

purification step displayed several peaks eluting in the region where azorhizobial Nod factors elute. These peaks were analyzed with FAB-MS and CID-MS, and in eight of them, LCOs were identified. As an example, the CID-MS spectrum of one of the major products is shown in Fig. 3D. The [M1H]1 pseudomolecular ion was at m/z 1244, and the fragment ions were at m/z 1023, 820, 617, and 414. This pattern corresponded to a pentasaccharide of GlcNAc with, on the nonreducing end, a C16:0 and a methyl substitution. The CID-MS spectra of the other products were exactly the same (loss of 221, followed by three losses of 203 and with the ion derived from two GlcNAc losses as the most intense) except that the masses differed. This means that all the molecules consisted of five GlcNAc but that the substitutions at the nonreducing end differed. To obtain additional structural information, the fatty acids were determined by gas chromatography. The results, together with the data from the mass spectrometry, are summarized in Table I. The HPLC peaks corresponding to the major NodABCS products (with a C16:0, C18:1, or C18:0 fatty acid) and some of the minor products were accompanied with a slightly slower migrating peak with approximately the same intensity. Products in these peaks showed the same MS spectrum as the products in the preceding peak except that they had a mass that is 42 units higher. This might indicate that these products were NodABCS metabolites, O-acetylated at the nonreducing end, a substitution that might be the result of an endogenous O-acetyl transferase. From the analysis of the water-soluble products of the in vivo labeling experiments with the strains DH5a(pUCNABC) and DH5a(pUCNABCS), it is clear that the NodBC and NodBCS metabolites, respectively, were still present, even after overnight incubation (Fig. 2A, lanes 6 and 8). The in vivo production of the NodBCS and NodABCS products was followed in function of time. DH5a(pUCNABCS) cells were labeled with [14C]GlcNAc, and, at regular time intervals, samples were analyzed. Both the NodBCS and NodABCS metabolites increased equally before reaching a maximum level (Fig. 6) at which they remained constant, even after 24 h of incubation (data not shown). The accumulation of NodBCS metabolites was not due to a stop in Nod metabolite synthesis. A control experiment, in which cells were incubated with [14C]acetate or [14C]GlcNAc at the time point where the NodBCS metabolites reached their constant level, showed that Nod metabolite synthesis was still going on (data not shown).

DISCUSSION

nod genes of rhizobia code for proteins that are involved in the biosynthesis of lipo-chitooligosaccharide Nod factors. Azorhizobial nodABCS genes were introduced in various combinations into E. coli strain DH5a, and we found that nodABC genes are necessary and sufficient to produce the Nod factor core. In the case of the nodABCS construct, the LCOs are pentamers acylated at the nonreducing end with fatty acids varying in length from C14:0 to C18:0. The MS data, together with the [3H]methionine in vivo labeling studies, further show that the NodABCS metabolites are methylated at the nonreducing end, except for two minor compounds (see Table I). This confirms that NodS is a methyl transferase involved in Nod factor methylation (9, 10). Data obtained from analysis of the Nod factor intermediates produced by E. coli strains expressing nodC, nodBC, nodBCS, nodABC, and nodABCS are in agreement with the described in vitro enzymatic activities of the proteins encoded by these genes and show that these enzymes produce in vivo the same products as in vitro: NodC is a N-acetylglucosaminyl transferase synthesizing chitooligosaccharides with polymerization degree from 2 to 5, NodB deacetylates these products at the nonreducing end, NodS methylates the deacetylated products, and NodA transfers an acyl chain to the free amine at the nonreducing end. The variety of acyl chains found in the NodABCS metabolites produced in an E. coli background shows that the NodA protein of A. caulinodans has low specificity for the chain length of the fatty acid. Bacterial polysaccharides are often synthesized on the lipid carrier undecaprenyl phosphate (12), and it was suggested that such lipids could be involved in Nod factor synthesis (1, 4). However, we found no evidence for this hypothesis. First, we never detected lipid-linked intermediates in vitro nor in vivo. Second, the antibiotics tunicamycin and bacitracin had no effect on the synthesis of Nod factors or NodC metabolites in vivo and in vitro in A. caulinodans and E. coli. Tunicamycin specifically inhibits the transfer of GlcNAc-1-phosphate from UDPGlcNAc to polyprenyl monophosphate acceptors. For example, in E. coli, the in vitro synthesis of GlcNAc-pyrophosphorylundecaprenol, an intermediate in the synthesis of enterobacterial common antigen, is totally inhibited by tunicamycin (13) as is the in vivo synthesis of enterobacterial common antigen (14). Bacitracin is a polypeptide antibiotic, inhibiting the dephosphorylation of polyprenol-pyrophosphate, thereby preventing the use of the lipid carrier (15).

29222


TABLE I Lipo-chitooligosaccharides produced by DH5a(pUCNABCS) *, major products; **, Nod metabolites also produced by ORS571(pRG70) (2).

a b c

Retention time

Mass [M1H]1

17.2 20.0 20.0 21.8 21.8 22.5 22.5 23.5 24.4 24.4 25.8 26.6

1230 1258 1230 1244* 1270*,** 1286* 1312 1258 1300 1286 1272*,** 1314*

Mass of fragments

400, 428, 400, 414, 440, 456, 482, 428, 470, 456, 442, 484,

603, 631, 603, 617, 643, 659, 685, 631, 673, 659, 645, 687,

806, 834, 806, 820, 846, 862, 888, 834, 876, 862, 848, 890,

1009 1037 1009 1023 1049 1065 1091 1037 1079 1065 1051 1093

DPa

Fatty acid b

V V V V V V V V V V V V

C16:0 C14:0 C15:0 C16:0 C18:1 C16:0 C18:1 C17:0 C17:0 C17:0 C18:0 C18:0

Substitutionsc

Methyl, Methyl Methyl Methyl Methyl, Methyl, Methyl Methyl, Acetyl Methyl Methyl,

acetyl

acetyl acetyl acetyl acetyl

Degree of polymerization, deduced from the number of losses of 201 mass units. Determined by gas chromatography. Deduced from the mass of the nonreducing end.

FIG. 6. Production of NodBCS and NodABCS metabolites by DH5a(pUCNABCS) in function of time. Nod metabolites were labeled in vivo using [14C]GlcNAc, extracted from cells at the indicated times, analyzed by TLC, and quantified with a PhosphorImager.

NodC is homologous to yeast chitin synthases (16) and hyaluronan synthase of group A streptococci (17). Also for these enzymes, no lipid-linked intermediates could be identified, and tunicamycin had no effect on the activity (18 –21). The in vitro time course (Figs. 4 and 5B) and the in vivo labeling of Nod metabolites in the strains DH5a(pUCNCS) and DH5a(pUCNBCS) (Fig. 5A) show the substrate-product relationship for the NodC-NodBC and NodBC-NodBCS products and that NodC, NodB, and NodS act in this order. The strain DH5a(pUCNABCS) produces both NodABCS and NodBCS metabolites, suggesting that also in the presence of NodA, NodS acts immediately after NodB, and NodA after NodS. The formation of LCOs in the strain DH5a(pUCNABCS) suggests a substrate-product relation for the NodBCS-NodABCS metabolites. A large pool of the NodBCS products, however, was not converted even though Nod factor synthesis was still proceeding. A possible explanation is that the NodC metabolites remain linked, covalently or noncovalently, to the NodC protein while the other proteins (NodB, NodS, and NodA) modify the protruding nonreducing end. Accumulation of NodBC or NodBCS metabolites in the strains DH5a(pUCNABC) or DH5a(pUCNABCS), respectively, would then be the outcome of insufficient NodA protein or acyl donor, resulting in the release from NodC in the cytoplasm where they cannot be acylated anymore. This would also explain why in vitro acylation of deacetylated chitooligosaccharides was unsuccessful in our hands and equally explains the inefficient in

vitro acylation reported by Atkinson et al. (7) (only 0.1% of the substrate was converted). The proposed mechanism requires oligosaccharide elongation from the nonreducing end toward the reducing end, in agreement with a model concerning the action of b-glycosyl transferases (22). The chain length of Nod factors would be determined by the length of the sugar chain at the moment that the nonreducing end becomes accessible for the other proteins. A consequence of this model is that modifications on the reducing end of the Nod factor core have to be introduced after the synthesis of the core is completed, while modifications at the nonreducing end can be introduced either before or after acylation. This fits with data of Schultze et al. (23) showing that the sulfate group at the reducing end of R. meliloti Nod factors is introduced after the core synthesis. The acetyl group present at the nonreducing end of the Nod factors of R. leguminosarum bv. viciae is introduced after deacetylation by NodB and before acylation by NodA (24). Also, our results concerning methylation by NodS (Ref. 10 and this work) are consistent with this. The E. coli strain expressing the nodABC genes provides a basis to analyze the functions of other nod genes in Nod factor modifications. Moreover, the strains DH5a(pUCNC) and DH5a(pUCNBC) can be used to determine where in the biochemical pathway a modification is introduced. The feasibility of such a strategy was shown in this work for the NodS protein. The role of NodU in carbamoylation of Nod factors was also confirmed making use of this approach.2 Acknowledgments—We are grateful to Marc Vancanneyt for gas chromatography analysis. We thank Koen Goethals and Manuel Fernańdez-Lo´pez for critical reading of the manuscript, and Martine De Cock, Christiane Germonprez, and Karel Spruyt for help with the manuscript and the figures. REFERENCES 1. Carlson, R. W., Price, N. P. J., and Stacey, G. (1994) Mol. Plant-Microbe Interact. 7, 684 – 695 2. Mergaert, P., Van Montagu, M., Prome´, J.-C., and Holsters, M. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 1551–1555 3. Geremia, R. A., Mergaert, P., Geelen, D., Van Montagu, M., and Holsters, M. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 2669 –2673 4. Spaink, H. P., Wijfjes, A. H. M., van der Drift, K. M. G. M., Haverkamp, J., Thomas-Oates, J. E., and Lugtenberg, B. J. J. (1994) Mol. Microbiol. 13, 821– 831 5. John, M., Ro¨hrig, H., Schmidt, J., Wieneke, U., and Schell, J. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 625– 629 6. Ro¨hrig, H., Schmidt, J., Wieneke, U., Kondorosi, E., Barlier, I., Schell, J., and John, M. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 3122–3126 7. Atkinson, E. M., Placic, M. M., Hindsgaul, O., and Long, S. R. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 8418 – 8422 8. Goethals, K., Gao, M., Tomekpe, K., Van Montagu, M., and Holsters, M. (1989) Mol. & Gen. Genet. 219, 289 –298 9. Geelen, D., Mergaert, P., Geremia, R. A., Goormachtig, S., Van Montagu, M., and Holsters, M. (1993) Mol. Microbiol. 9, 145–154 10. Geelen, D., Leyman, B., Mergaert, P., Klarskov, K., Van Montagu, M.,

Nod Factor Biosynthesis in A. caulinodans Geremia, R., and Holsters, M. (1995) Mol. Microbiol. 17, 387–397 11. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY 12. Bugg, T. D. H., and Brandish, P. E. (1994) FEMS Microbiol. Lett. 119, 255–262 13. Barr, K., and Rick, P. D. (1987) J. Biol. Chem. 262, 7142–7150 14. Rick, P. D., Mayer, H., Neumeyer, B. A., Wolski, S., and Bitter-Suermann, D. (1985) J. Bacteriol. 162, 494 –503 15. Siewert, G., and Strominger, J. L. (1967) Proc. Natl. Acad. Sci. U. S. A. 57, 767–773 16. Bulawa, C. E., and Wasco, W. (1991) Nature 353, 710 17. DeAngelis, P. L., Papaconstantinou, J., and Weigel, P. H. (1993) J. Biol. Chem. 268, 19181–19184

18. 19. 20. 21.

29223

Duran, A., and Cabib, E. (1978) J. Biol. Chem. 253, 4419 – 4425 Orlean, P. (1987) J. Biol. Chem. 262, 5732–5739 Stoolmiller, A. C., and Dorfman, A. (1969) J. Biol. Chem. 244, 236 –246 Sugahara, K., Schwartz, N. B., and Dorfman, A. (1979) J. Biol. Chem. 254, 6252– 6261 22. Saxena, I. M., Brown, R. M., Jr., Fevre, M., Geremia, R. A., and Henrissat, B. (1995) J. Bacteriol. 177, 1419 –1424 23. Schultze, M., Staehelin, C., Ro¨hrig, H., John, M., Schmidt, J., Kondorosi, E., Schell, J., and Kondorosi, A. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 2706 –2709 24. Bloemberg, G.V. (1995) Biochemical function of the host specificity-determining nodulation genes of Rhizobium leguminosarum. Ph.D. thesis, Rijksuniversiteit Leiden, The Netherlands