Dictyostelium discoideum Fatty-acyl Amidase II Has Deacylase Activity on Rhizobium Nodulation Factors*

Dictyostelium discoideum (Amoebidae) secretes cell-lysing enzymes: esterases, amidases, and glycosylases, many of which degrade soil bacteria to provide a source of nutrients. Two of these enzymes, fatty-acyl amidases FAA I and FAA II, act sequentially on the N -linked long chain acyl groups of lipid A, the lipid anchor of Gram-negative bacterial lipopolysaccharide. FAA I selectively hydrolyzes the 3-hydroxymyristoyl group N -linked to the proximal glucosamine residue of de- O -acylated lipid A. Substrate specificity for FAA II is less selective, but does require prior de- N -acylation of the proximal sugar, i.e. bis- N -acylated lipid A is not a substrate. We have synthesized a 14 C-labeled substrate analog for FAA II and used this in a novel assay to monitor its purification. Inhibitory studies indicate that FAA II is not a serine protease, but may have a catalytic mechanism similar to metalloprotein de- N -acetylases such as LpxC. Interestingly, rhizobial Nod factor signal oligosaccha-rides that induce root nodules on leguminous plants have many of the structural requirements for substrate recognition by FAA II. In vitro evidence indicates that Rhizobium fredii Nod factors are selectively de- N -acy-lated by purified FAA II, suggesting that the enzyme may reduce the N 2 -fixing efficiency of Rhizobium

Dictyostelium discoideum secretes hydrolytic enzymes that degrade bacterial cell walls, enabling it to utilize soil-borne bacteria as a source of nutrition. Many of these enzymes have been implicated in the degradation of bacterial lipopolysaccharide, which is found exclusively in the cell walls of Gram-negative bacteria (1,2). Two enzymes, fatty-acyl amidases FAA 1 I and FAA II, are of particular interest and play a concerted role in the deacylation of lipid A (3), a molecule that anchors lipopolysaccharide to the bacterial outer membrane (Fig. 1). FAA I is highly specific in its substrate requirements and selectively hydrolyzes N-linked fatty acids attached to the proximal sugar of de-O-acylated lipid A (1). This monodeacylated lipid A is then further hydrolyzed by the action of FAA II, releasing fully deacylated lipid A oligosaccharide (1,3). FAA II is less specific than FAA I, but requires that the lipid A substrate is mono-Nacylated only on the distal sugar and that the fatty acid is longer than three carbons and 2-N-linked to a glucosaminyl residue. There is no requirement for the fatty acid to be 3-hydroxylated (as it is in lipid A) or for the glucosaminyl moiety to be phosphorylated or glycosidically linked (1). Consequently, the minimum structures for substrate recognition by FAA II are 2-N-acylamino-2-deoxy-D-glucopyranoses.
Nitrogen-fixing bacteria of the genus Rhizobium secrete Nod factor signal molecules that induce symbiotic root nodules on certain leguminous plants (4). Significantly, these Nod factor lipochito-oligosaccharides have many of the structural requirements for substrate recognition by FAA II (see Fig. 3). The long chain fatty acid moiety (typically C 16 -C 18 ) is 2-N-linked to the distal glucosamine residue of Nod factors and, unlike bis-Nacylated lipid A, is not blocked by further acylation of the oligosaccharide core. Nod factor formation proceeds by NodCcatalyzed biosynthesis of chito-oligosaccharides, which are subsequently modified by the presence of host-specificity nodulation functions. The N-acyl chain is a common feature of all Nod factor structures described to date and is essential for their biological activity (5,6). However, certain Nod factor structures are modified at the N-acylated terminus by conjugated double bonds (7,8) or by N-methylation of the nonreducing residue (9 -12). This N-methylation is catalyzed by NodS protein that is present in Rhizobium sp. NGR234, Rhizobium etli, and certain Bradyrhizobium species, but not in Rhizobium fredii USDA257 (13). Potentially, N-methylation or the presence of conjugated polyunsaturation may stabilize the acyl amide bond and therefore protect Nod factors against degradation by FAA II.
We have synthesized N-palmitoyl [1-14 C]glucosamine (GlcN-C 16 ) as a synthetic substrate and found it to be readily de-Nacylated by the action of FAA II. This has been utilized in a novel assay for FAA II and used to purify the protein to homogeneity from the culture medium of a Dictyostelium nagA knockout (14). Susceptibility to various inhibitors suggests that FAA II has mechanistic similarities to zinc-containing acetamidases, such as the 3-O-acyl-GlcNAc de-N-acetylase LpxC (15), rather than to serine proteases. The activity of purified FAA II has also been assessed on metabolically radiolabeled Nod factors from R. fredii USDA257, and the results indicate that these also become de-N-acylated. Since Nod factors lacking the N-acyl chain are biologically inactive, the presence of FAA II in soil may reduce the ability of Rhizobium to nodulate host leguminous plants. Interestingly, Nod factors from a transconjugate of USDA257 expressing the nodS N-methyltransferase gene from Rhizobium sp. NGR234 are found to be more resistant to FAA II, suggesting that N-methylation may protect Nod factors against deacylation by Dictyostelium in soil. C]Glucosamine (0.5 l, 50 nCi) was spotted at the origin on a normal-phase Silica Gel 60 thin-layer chromatography plate, dried under a stream of warm air, and overspotted with palmitoyl chloride (2 l). After an 18-h reaction at room temperature, the plate was developed with chloroform, methanol, and aqueous ammonium hydroxide (50:40:5 by volume). Untreated [1-14 C]glucosamine was run in a control lane. The plate was dried until free of solvent and autoradiographed for 2 days. Radiolabeled N-palmitoylglucosamine was recovered from the plates by extraction with methanol. Following scintillation counting, aliquots were evaporated to dryness in Eppendorf tubes and stored at Ϫ20°C.

Materials
Isolation and Purification of FAA II-Culture supernatant (14 liters) from D. discoideum strain Ax-3 was concentrated by passage through a 30-kDa cutoff Amicon filter, and the concentrated supernatant (1 liter) was stirred overnight with DEAE-cellulose (50 g, 4°C). The ion exchanger was recovered by filtration through a cotton pad, washed with NaCl (150 mM) in cold Buffer A (125 ml), and eluted with NaCl (1 M) in Buffer A (187 ml). Following precipitation with ammonium sulfate (70%), and centrifugation (15,000 ϫ g, 15 min, 4°C), the protein pellet was redissolved in Buffer B (3 ml) and dialyzed overnight against Buffer B. It was further fractionated on a DEAE-cellulose column (20 ϫ 1 cm) as follows: 1) washed with Buffer B (5 bed volumes) and 2) eluted with a gradient of NaCl (150 -500 mM) in Buffer B. Fractions were assayed for protein using the BCA assay (16) and for FAA II activity as described below. Active fractions were stored at Ϫ20°C prior to purification by high performance liquid chromatography.
High Performance Liquid Chromatography (HPLC)-The HPLC system consisted of two Waters 512 pumps, a Waters 486 UV detector, and a U6K injector. Reverse-phase chromatography was achieved on a Brownlee Spheri-5 RP-18 column (250 mm, 5-m particle size) in series with a Waters RP-18 column (150 mm, 5-m particle size). A gradient from 0.01% trifluoroacetic acid in 10% aqueous acetonitrile to 0.05% trifluoroacetic acid in 100% acetonitrile was run for 30 min at 0.6 ml/min. Gel-filtration HPLC was done on a Phenomenex Biosep-Sec-S-3000 column (300 ϫ 7.8 mm) protected by a Phenomenex Biosep-Sec-S guard column (7.5 ϫ 7.8 mm) using a mobile phase of 50 mM sodium phosphate buffer (pH 6.8; 1 ml/min). In either case, detection was achieved by absorbance at 210 nm. A standard protein ladder was run as a molecular mass marker. Fractions were collected from the columns and assayed for protein and FAA II activity. SDS-polyacrylamide gel electrophoresis was run as per Laemmli (17).
Enzyme Assay and Inhibition Study of the Amidase FAA II-Enzyme assays consisted of N-palmitoyl[U-14 C]glucosamine (a minimum of 10 nCi/assay) and protein (1-10 mg/ml) in 0.2% Triton X-100 in Buffer B at a final volume of 100 l. The reaction was allowed to proceed to completion (18 h, 26°C), after which it was stopped by the addition of aqueous NaCl (10% (w/v), 100 l). Unreacted substrate was extracted by partitioning with ethyl acetate/hexane (90:10 (v/v), 300 l). Following microcentrifugation, radioactivity was assessed in the upper solvent layer (200 l) and lower aqueous layer (100 l) by scintillation counting. A minimum of three replications was analyzed per assay. Controls were identical, but lacked protein exudate. Stock solutions of the inhibitors phenylmethylsulfonyl fluoride (50 mM) and EDTA (1 M) were prepared in acetonitrile and Buffer B, respectively, and, when used, were added prior to the protein addition.
The FAA II assay was modified for testing on Nod factors so that activity could be monitored by TLC/autoradiography. Metabolically radiolabeled Nod factors (typically 30 nCi) were incubated with FAA II (10 mg/ml, 18 h, 30°C) in Buffer B (30 l) containing 0.2% Triton X-100. Control reactions lacked the enzyme. The reaction was stopped by the addition of methanol (50 l), and the precipitated protein was pelleted by brief centrifugation (15,000 rpm, 15 min). Aliquots of the methanolic supernatant (30 l) were separated on silica TLC plates eluted with butanol/ethanol/water (5:3:2 by volume). Plates were air-dried and exposed to x-ray film for 6 days.
Preparation and Testing of Metabolically 14 C-Labeled Nod Factors-Rhizobia were grown aerobically at 30°C on rhizobial minimal medium (RMM) in the presence of succinate (12 mM) and glutamate (6 mM) as described previously (11). Cultures (10 ml) were radiolabeled by the inclusion of [1-14 C]glucosamine (40 -60 mCi/mmol, 1 Ci/ml) at the start of the growth period. After 4 h of growth (A 550 ϭ 0.3), the nod genes were induced by the addition of the species-specific flavonoid apigenin (5 ϫ 10 Ϫ4 M). After 18 h, the bacteria were removed by centrifugation (10,000 ϫ g, 4°C), and radiolabeled Nod factors were recovered from the culture supernatant by solid-phase extraction with C 18 -functionalized silica (Waters) (18). After washing with water (5 ml), radiolabeled metabolites were eluted with methanol and evaporated to dryness.
Alternatively, Nod factors were metabolically labeled from precursor [ 14 C]acetate. Cultures (10 ml) were grown to logarithmic phase (A 550 ϭ 1.0) as described above and divided, and one-half was induced with apigenin (5 ϫ 10 Ϫ4 M). After diluting to 10 ml with fresh RMM and an additional 2 h of growth, the cells were collected by centrifugation, washed once, and resuspended in 1 ml of RMM without the carbon source. Cells were incubated for 10 min with sodium [2-14 C]acetate (54 mCi/ml, 10 Ci) and then diluted to 50 ml with RMM containing succinate (12 mM) and glutamate (6 mM). After 2 days further growth, the radiolabeled Nod factors were isolated as described above.

Substrate Specificity and Development of an Enzyme
Assay-We prepared and purified, by thin-layer chromatography with autoradiographic detection (Fig. 2), a radiolabeled synthetic substrate for FAA II, N-palmitoyl[1-14 C]glucosamine (Fig. 3). Conditions were optimized for two-phase partitioning of N-palmitoyl [1-14 C]glucosamine from the free sugar [1-14 C]glucosamine, making use of ethyl acetate/hexane (90:10, v/v) and aqueous sodium chloride (10%, w/v). This allowed selective recovery of the radiolabeled monosaccharide in the aqueous layer. Thus, after treatment of N-palmitoyl[1-14 C]glucosamine with FAA II in aqueous Buffer B and the addition of sodium chloride (to 10%, w/v), selective counting of the aqueous and organic phases is an indication of the reaction progress. The salt also served the dual purpose of stopping the reaction. Solvent extraction of radiolabeled GlcN-C 16 was optimized such that 98% of the radioactivity from the enzyme-free control partitioned into the solvent phase (Table I) tracts was monitored, allowing for complete purification and a mechanistic study of FAA II.
Purification and Mechanistic Studies of FAA II-Purification was undertaken on a secreted fraction from the culture supernatant of Dictyostelium strain Ax-3 after passage through a 30-kDa cutoff filter. Activity was monitored by the two-phase partitioning assay. Enzyme assays were typically run at 10 mg/ml total protein, giving a 56.3% decrease in counts in the organic phase relative to controls after a 18-h reaction time. A concomitant increase in counts was recorded in the aqueous extraction buffer. Lower concentrations of FAA II or shorter reaction times resulted in decreased activity. Final purity of the enzyme was ascertained by SDS-polyacrylamide gel electrophoresis and reverse-phase HPLC (Fig. 4). Verret et al. (2) were unable to fully purify FAA II free from acetylhexosaminidase A (NagA), but estimated its relative mass to be in the range of 60 -80 kDa. Here the molecular mass of FAA II was determined by gel electrophoresis and gel-filtration HPLC. The protein eluted on gels with a molecular mass of 66,000 (Fig. 4). By gel filtration, the same purified fraction eluted as a single peak, but in this case, the relative molecular mass was estimated at 32,985 Da. FAA II may be mechanistically similar to serine proteases or metalloproteases in that amide bonds are being hydrolyzed. To examine this, we assayed purified FAA II in the presence of two known protease inhibitors (Table I). Stock solutions of phenylmethylsulfonyl fluoride were prepared at 50 mM in either methanol or acetonitrile and used at the concentrations shown in Table I. EDTA was prepared in Buffer B (292 mg/ml) and used at a final concentration of 50 mM. As Table I indicates, these inhibitors had no significant effect on the activity of FAA II under the conditions tested, and results were within Ϯ4.1% of the non-inhibited FAA II reaction.
FAA II Selectively Deacylates Rhizobial Nod Factors-Rhizobial Nod factors were radiolabeled metabolically by culturing rhizobial strains on minimal medium in the presence of pre-cursor [1-14 C]glucosamine and apigenin, an appropriate flavonoid nod gene inducer (18). Alternatively, optimal incorporation of radiolabel from acetate was attained by incubation with sodium [2-14 C]acetate as the sole carbon source prior to chasing with succinate (12 mM) and glutamate (6 mM). The secreted hydrophobic fractions containing the Nod factors were separated from excess radiolabeled precursor and other labeled polar metabolites on a reverse-phase cartridge. The induction of Nod factors, as assessed by TLC/autoradiography, was clearly dependent on the presence of apigenin (Fig. 5). Induci-   16 Assays were performed as described under "Experimental Procedures." Unreacted substrate was extracted by partitioning between aqueous NaCl (10%, w/v) in Buffer B (200 l) and ethyl acetate/hexane (90:10 (v/v), 300 l). Following microcentrifugation, radioactivity was assessed by scintillation counting, with the de-N-acylated product partitioning into the aqueous layer. Control samples lacked protein. Inhibitors (phenylmethylsulfonyl fluoride and EDTA) were introduced at the final concentrations indicated. The FAA II assay was modified so that activity on Nod factors could be monitored by TLC/autoradiography. Hydrophobic extracts of the conditioned culture medium from apigenin-induced or -uninduced R. fredii cultures were treated in parallel with FAA II prior to chromatographic separation (Fig.  5). Comparable reactions lacking the FAA II enzyme served as controls. Treatment of [ 14 C]acetate-labeled Nod factors with FAA II for 18 h resulted in an attenuation of the labeling (Fig.  5, A and B), suggesting that either the hydrolyzed acyl group or the unhydrolyzed Nod factor was sequestered with the amidase. This was confirmed by the presence of radioactivity in the protein pellet. Extracts from uninduced cultures were unaffected by this treatment. Similar experiments on [ 14 C]glucosamine-labeled Nod factors resulted in a chromatographic mobility shift, but without attenuation of intensity (Fig. 5C). These data are consistent with metabolic incorporation of [ 14 C]acetate into the Nod factor acyl chain and of [ 14 C]glucosamine into the oligosaccharide backbone. Following deacylation, radiolabel is lost from [ 14 C]acetate-labeled Nod factors or retained but chromatographically shifted for [ 14 C]glucosamine-labeled Nod factors.
Wild-type R. fredii USDA257 produces fucosylated and Nvaccenoylated penta-, tetra-, and trisaccharide Nod factors that induce nodules on soybean (19). Other Rhizobium species produce N-vaccenoylated Nod factors on which the acylated nitrogen is also N-methylated, such that the terminal nonreducing residue is N-vaccenoyl-N-methyl-2-deoxy-D-glucosamine. The presence of the N-methyl group has been attributed to the activity of a rhizobial inducible N-methyltransferase (NodS) (20). To assess the susceptibility of N-methylated Nod factors to FAA II, we isolated a metabolically labeled Nod factor fraction from a USDA257 transconjugate expressing the nodS region from Rhizobium sp. NGR234. The Nod factors from this strain were shown to be N-methylated by the extension of the symbiotic host range to include the tropical legume Leucaena, which is selectively responsive to N-methylated Nod factors (13), and by the flavonoid-inducible incorporation of metabolic label from [methyl-14 C]methionine. 2 Apigenin-inducible, [ 14 C]glucosamine-labeled Nod factors were obtained from this strain as observed by TLC/autoradiography (Fig. 6). However, unlike the wild-type R. fredii Nod factor, the majority of the N-methylated factor was unaffected by the FAA II treatment 2 A. E. Tobin and N. P. J. Price, manuscript in preparation.

FIG. 4.
Purification of fatty-acyl amidase II. FAA II was purified from D. discoideum culture exudate by 1) passage through a 30-kDa cutoff filter, 2) DEAE ion-exchange chromatography (pH 7.8), 3) salting out with ammonium sulfate (70%), and 4) DEAE ion-exchange chromatography (pH 4.0). Fractions were assayed for protein using the BCA assay and for FAA II activity as described under "Experimental Procedures." Active fractions were stored at Ϫ20°C prior to purification by high performance liquid chromatography. A, SDS-polyacrylamide gel stained with Coomassie Blue. Left lane, FAA II (indicated by an arrow); right lane, molecular mass markers (66 and 48.5 kDa indicated by arrows). B, reverse-phase HPLC trace of FAA II after separation on a Spheri-5 C 18 analytical column with a linear gradient (10 -100% acetonitrile and 0.05% trifluoroacetic acid). Detection was by absorbance at 220 nm and by BCA protein assay.
FIG. 5. Activity of FAA II on metabolically radiolabeled Nod factors (lipo-oligosaccharides) from R. fredii. Nod factors were isolated from R. fredii USDA257 grown on RMM medium with appropriate radiolabeled precursors. The Nod factor fractions were isolated on a C 18 reverse-phase cartridge, washed free of excess radiolabel, and eluted with methanol. Amidase reactions were started by the addition of FAA II in Buffer B (10 mg/ml, 100 l) plus Triton X-100 (0.2%). Following incubation (18 h, 30°C), an equivalent volume of methanol was added to stop the reaction, and samples were centrifuged (10,000 ϫ g, 2 min) and spotted onto Silica Gel 60 TLC plates. Plates were eluted with butanol/ethanol/water (5:3:2 by volume), air-dried, and autoradiographed for 5 days. and was observed on plates to co-migrate with the untreated control (Fig. 6). DISCUSSION Fatty-acyl amidases play a concerted role in the deacylation of bacterial lipid A (Fig. 1) by selective removal of the N-linked fatty acid chains. The presumed natural substrate for FAA II is generated by degradation of de-O-acylated lipid IV A by FAA I to form 4-phosphoryl-N-␤-hydroxymyristoyl-D-glucosaminyl-␤-1,6-glucosamine 1-phosphate. Removal of the O-acyl chains and the 2-deoxyketooctulosonic acid residues is presumed to occur prior to this due to the action of other secreted hydrolases (3). Interestingly, neither amidase requires the C-1 or C-4Ј phosphate group for substrate recognition, and although FAA II requires a deacylated proximal sugar, the positive charge on the free 2-amino group is not obligatory since 2-N-formylation does not significantly inhibit FAA II (1). For many natural endotoxins, the distal fatty acid amide is ␤-3-hydroxylated, although the 3-hydroxy group is also unnecessary for substrate recognition by FAA II. Furthermore, the ␤-1,6-glycosidic linkage is also nonessential, indicating that N-acyl-D-glucosaminyl monosaccharides are appropriate as alternate substrates.
Based on this information, we synthesized N-palmitoyl-D- [1-14 C]glucosamine as a radiolabeled substrate for FAA II. This was prepared by selective acylation of D-[1-14 C]glucosamine HCl with palmitoyl chloride and was subsequently used in a novel radioenzyme assay to purify FAA II to homogeneity. The radiolabeled substrate and the depalmitoylated product were effectively separated by solvent partitioning (Table I). Using this enzyme assay, a five-step purification procedure afforded purified FAA II. Partially purified fractions also contained secreted ␤-N-acetylhexosaminidase A, the product of the nagA gene. Because ␤-N-acetylhexosaminidase A might degrade ␤-1,4-glucosidic linkages, such as are found in Nod factors, it was important that this activity was removed, particularly since the NagA protein co-migrated with FAA II both on reverse-phase HPLC columns and by SDS gel electrophoresis. To circumvent these problems, we made use of a nagA-deleted strain of Dictyostelium strain Ax-3, HL101 (14), such that subsequent protein purifications were made from the culture medium of this knockout strain. Consequently and in addition to earlier work, we were able to purify FAA II to homogeneity free from ␤-N-acetylhexosaminidase A. Interestingly, a large discrepancy was found for the relative molecular mass of FAA II as obtained by SDS gel electrophoresis (66 kDa) or sizeexclusion HPLC (33 kDa). This may be attributed to posttranslational modification of FAA II particularly if polyanionic groups such as sulfate or methyl phosphate predominate, as has been found for other secreted dictyostelial proteins (21,22).
The enzyme mechanism of FAA II has not been studied previously, although the activity is suggestive of either a serine (or cysteine) protease or a zinc-containing metalloprotein. Serine proteases have a bimolecular mechanism that proceeds via a covalent acyl-enzyme intermediate. Thus, for FAA II, this would suggest a two-step displacement involving the sugar moiety followed by the acyl group. By contrast, a one-step metalloprotein mechanism, such as for carboxypeptidase A, would initially give rise to a noncovalent zinc-acyl transition state in which the acyl carbonyl oxygen is coordinated to the zinc atom. Significantly, the protease activity of carboxypeptidase A is selective for the N-linked terminal amino acid, analogous to the sugar moiety of GlcN-C 16 , but is nonselective with regard to the peptide chain. To test these mechanisms, we assayed FAA II in the presence of various protease inhibitors. FAA II was unaffected by either a serine protease inhibitor (phenylmethylsulfonyl fluoride) or a sequestering agent (EDTA). The latter was surprising in view of previous work involving the bacterial lipopolysaccharide deacetylase LpxC (15). LpxC is a lipid A biosynthetic enzyme that de-N-acetylates the substrate UDP-3-O-acyl-GlcNAc in a reaction analogous to FAA II amide hydrolysis. LpxC is a zinc-containing metalloprotein and has been assigned to a unique class of metalloamidases. Although LpxC from Pseudomonas aeruginosa is sensitive to inhibition by EDTA, the analogous protein from Escherichia coli is unaffected presumably because the zinc atom is buried within the protein (15). FAA II is an extracellular protein that most probably requires additional stability to maintain a folded conformation. It too may have an inaccessible buried metallo-center so that enzymatic activity is retained in soil.
We have shown that Nod factors are selectively de-N-acylated by the action of the fatty-acyl amidase FAA II. Since non-acylated Nod factors lack nodule-promoting biological activity, deacylation may well have an effect upon the ability of rhizobia to nodulate host plants. Moreover, we propose that Nod factors that are modified by the presence of conjugated double bonds (such as those from Rhizobium meliloti and Rhizobium leguminosarum) or by N-methylation of the nonreducing residue (such as, for example, Rhizobium NGR234) may be altered in their sensitivity to enzymatic de-N-acylation by FAA II. Consequently, these modifications may protect Nod factors against degradation by D. discoideum in soil. Truchet et al. (23) have shown that nodulation efficiency is concentration-dependent and that nodule induction does not occur below a threshold Nod factor concentration of ϳ10 Ϫ7 M, even though a legume root hair deformation response is observed at far lower concentrations. FAA II de-N-acylation of Nod factors may therefore reduce the efficiency of legume nodulation by reducing the concentration of available biologically active Nod factors. Since Nod factors and FAA II are both secreted by soil-dwelling Rhizobium and Dictyostelium species, respectively, Nod factor degradation may also occur in soils by this mechanism, decreasing the efficiency of nitrogen-fixing Rhizobium-legume symbioses.