Rhizobium leguminosarum bv. trifolii produces Lipo-chitin Oligosaccharides with nodE -dependent Highly Unsaturated Fatty Acyl Moieties AN ELECTROSPRAY IONIZATION AND COLLISION-INDUCED DISSOCIATION TANDEM MASS SPECTROMETRIC STUDY*

The lipo-chitin oligosaccharides (LCO) or nodulation factors synthesized by Rhizobium leguminosarum bv. trifolii were analyzed using positive mode fast atom bombardment and positive and negative mode electrospray ionization mass spectrometry. From their mass spectrometric behavior it is clearly possible to distin-guish between the [M (cid:49) Na] (cid:49) pseudomolecular ion of the nodE -independent molecule IV(C18:1,Ac) and the [M (cid:49) H] (cid:49) pseudomolecular ion of the nodE -dependent molecule IV(C20:4,Ac), although they both have the same mass value. The results unequivocally show that the bacterial strain investigated produces nodE -dependent LCOs with highly unsaturated fatty acyl moieties. We further demonstrate that the interpretation of the mass spectrometric data by Philip-Hollingsworth et al . (Philip-Hollingsworth,

The symbiosis between bacteria belonging to the genera Rhizobium, Bradyrhizobium, and Azorhizobium and their leguminous host plants results in the formation of nitrogen-fixing root nodules. This symbiosis is regarded as being highly host-specific. Some bacterial strains, for example Rhizobium sp. NGR234 (1), are able to nodulate a broad range of host plants, while other strains are able to nodulate only a few plant species. Examples of rhizobia with a narrow host range are Rhizobium leguminosarum (R. leguminosarum) bv. trifolii, which nodulates plants of the genus Trifolium and the closely related R. leguminosarum bv. viciae which nodulates plants of the genera Lathyrus, Lens, Pisum, and Vicia.
It is generally recognized that the host specificity is mediated by signal molecules from both the plant and the bacterium. Flavonoids secreted by the plant are recognized by the guest bacteria resulting in the transcription of the nodulation genes (2,3). These genes encode the nodulation enzymes involved in the biosynthesis of lipo-chitin oligosaccharide (LCO) 1 signal molecules, also called Nod metabolites or Nod factors. In general, LCOs consist of a chitin backbone of three to six ␤134linked N-acetylglucosamine (GlcNAc) residues, N-acylated on the nonreducing terminal residue and bearing various biovarspecific modifications on both terminal residues (see e.g. Refs. 4 -6). Although the mechanism is not yet clear, the LCOs are involved in mediating the host-specific biological activity on the plant (4,6).
In a previous article (7) we described the analysis of LCOs produced by R. leguminosarum bv. trifolii. Several analytical techniques, such as reversed phase high performance liquid chromatography (reversed phase-HPLC) with photodiode array detection, fast atom bombardment mass spectrometry (FAB-MS), collision-induced dissociation tandem mass spectrometry (CID-MS-MS), and chemical modifications prior to mass spectrometry were applied. It was demonstrated that a mixture of LCOs is produced, with a chitin-type backbone of three to five GlcNAc residues, and bearing an O-acetyl group on the nonreducing terminal residue. The fatty acyl moieties vary from the common C16:0, C16:1, C18:0, C18:1, and C20:1 to the specific nodE-dependent highly unsaturated C18:3, C20:2, C20:3, and C20:4. It was postulated that these highly unsaturated fatty acyl moieties, although produced in a minor quantity compared to the nodE-independent common LCOs, determine the host specificity of R. leguminosarum bv. trifolii.
In a recent study, Philip-Hollingsworth et al. (8) rejected our assignment of highly unsaturated LCOs and suggested that they have "critically evaluated the proposed function of the nodE gene by performing detailed structural analysis" of LCOs produced by R. leguminosarum bv. trifolii. Based on data obtained from techniques such as electrospray ionization (ESI) mass spectrometry and gas chromatography coupled with mass spectrometry, no LCOs with highly unsaturated fatty acyl moieties could be detected in the cell membrane extracts. Furthermore, no differences were found in the spectra of LCOs produced by a wild-type strain, a strain lacking the nodE gene and a strain overexpressing the nodE gene. Since the two conflicting studies were carried out using different strains of R. leguminosarum bv. trifolii, and employing different methods of culturing and extracting the LCOs from the cell cultures, the fact that the authors of Ref. 8 were unable to detect LCOs with highly unsaturated fatty acyl moieties certainly does not dem-* 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. 1 The abbreviations used are: LCO, lipo-chitin oligosaccharides; HPLC, high performance liquid chromatography; FAB-MS, fast atom bombardment mass spectrometry; CID-MS-MS, collision-induced dissociation tandem mass spectrometry; ESI, electrospray ionization.
onstrate that these molecules cannot be produced by the bacteria under any conditions. To account for the absence of LCOs with highly unsaturated fatty acyl moieties, Philip-Hollingsworth et al. (8) suggest that the ions in the mass spectra assigned as nodE-dependent LCOs by Spaink et al. (7) were actually due to sodium adducts of the FIG. 1. Photodiode array UV spectrogram of reversed phase HPLC eluent of LCOs derived from strain LPR5045.pRtRF101. The absorbance of the late eluting fractions is not shown, since it is obscured by the peaks around fraction 35. For this HPLC analysis an extract (obtained after prepurification as described previously (7)) derived from 200 ml of culture was used. nodE-dependent Lipo-chitin Oligosaccharides of R. trifolii major nodE-independent LCOs. This reassignment was based on the fact that "no corresponding molecular ions were found in the negative mode electrospray ionization mass spectra of the same samples" (8). We present data that prove that it is not possible to confuse an [M ϩ H] ϩ with an [M ϩ Na] ϩ pseudomolecular ion.
In this paper we describe additional experiments we have carried out to reinforce our structural assignment of the structures of the nodE-dependent LCOs produced by R. leguminosarum bv. trifolii. For this purpose the LCOs were submitted to both positive and negative mode ESI-MS and to FAB-MS and CID-MS-MS experiments. The results clearly support our earlier findings (7) that R. leguminosarum bv. trifolii indeed produces LCOs with highly unsaturated fatty acyl moieties.
Electrospray Mass Spectrometry-Positive and negative ion mode electrospray mass spectra were obtained using a VG Platform II single quadrupole mass spectrometer. Aliquots of 10 l of the pooled fractions were infused into a mobile phase of acetonitrile/water (50: 50, v/v) and introduced into the electrospray source at a flow rate of  nodE-dependent Lipo-chitin Oligosaccharides of R. trifolii 5 l/min. In the positive mode spectra were scanned at a speed of 10 s for m/z 800-1400, with a cone voltage of 90 V. In the negative mode spectra were scanned at a speed of 10 s for m/z 100-1500, using a cone voltage of Ϫ40 V. All spectra were recorded and processed using the MassLynx 2.0 software. Fast Atom Bombardment and Collision-induced Dissociation Tandem Mass Spectrometry-FAB-mass spectra were obtained in the positive ion mode using a JEOL JMS-SX/SX102A tandem mass spectrometer using 10 kV accelerating voltage. The FAB gun was operated at 6 kV with an emission current of 10 mA using xenon as the bombarding gas. Spectra were scanned at a speed of 30 s for m/z 100-2000 and recorded and processed on a Hewlett-Packard HP9000 series data system using the JEOL Complement software. Collision-induced dissociation tandem mass spectra were obtained on the same instrument, using helium as the collision gas (in the third field free region) at a pressure sufficient to reduce the parent ion to one third of its original intensity. Aliquots of 2 l of the pooled fractions were loaded into a matrix of either thioglycerol or m-nitrobenzyl alcohol dosed with sodium.

Isolation and Separation of Lipo-chitin Oligosaccharides-
The HPLC diode-array profile (Fig. 1) clearly shows the presence of compounds having UV absorption maxima at 260, 303, or 330 nm, which we have previously assigned as corresponding to LCOs containing highly unsaturated fatty acyl moieties. These absorption maxima are indicative of a conjugated -system with two, three, or four double bonds, respectively, in conjugation with a carbonyl group (9, 10). The UV absorption spectra of synthetically produced fatty acyl moieties, which contain three or four such double bonds, support this assignment (11). 2 It should be noted that the species absorbing with nodE-dependent Lipo-chitin Oligosaccharides of R. trifolii 22567 a maximum of 260, 303, and 330 nm give UV absorbances of sufficient intensity to be detectable with any commercially available HPLC diode array detection system. The failure of Philip-Hollingsworth et al. (8) to observe such UV-absorbing peaks in their extracts can be explained in several ways. First, it has been observed that different growth conditions can change the relative quantities of the different types of LCOs produced by one bacterial strain (10,13,14). It is therefore quite possible that the strain studied by Philip-Hollingsworth et al. (8) does not produce LCOs with UV absorption maxima at 303 or 330 nm, due to their employing growth conditions that are unfavorable for the production of highly unsaturated fatty acids by NodE. Second, the solvent conditions chosen by Philip-Hollingsworth et al. (8) might not be suitable for the extraction of the C20 acyl containing LCOs; in our hands their conditions are not suitable for dissolving the classes of LCOs eluting later than 22 min in our HPLC chromatogram. Third, since the wild-type strain ANU843 has a rather unstable symbiotic phenotype, 3 the strain used by Philip-Hollingsworth et al. (8) might have suffered from mutations affecting nodE expression. However, regardless of the cause for the failure of Philip-Hollingsworth et al. (8) to detect specific LCOs, it should be noted that their results do not imply that the production and detection of highly unsaturated LCOs is not possible using a different strain, different growth conditions, and/or different purification methods.
ESI-MS-The so-called soft ionization mass spectrometric techniques, such as electrospray or fast atom bombardment ionization, generate ions in the positive mode by the addition of a cation to the sample molecule. In most cases the molecules are protonated, yielding an [M ϩ H] ϩ pseudomolecular ion, which is detected at an m/z value 1 mass unit higher than the molecular mass of the sample molecule. Because of the use of an aqueous mobile phase to transport the sample molecules to the electrospray interface, the sample solution is exposed to sodium ions, which leads to the formation of an [M ϩ Na] ϩ pseudomolecular ion. In the negative mode, pseudomolecular ions are generated either by the loss of a proton or by the addition of an anion.   Table I). In the negative mode, abundant clusters of three pseudomolecular ions were observed in the mass spectrum (Fig. 2, lower panel, and Table I Table I) at m/z 1119, 1139, and 1141 corresponding to [M ϩ Na] ϩ pseudomolecular ions for IV(C18:0,Ac), IV(C20: 4,Ac), and IV(C20:3,Ac). The corresponding pseudomolecular ions in the negative mode (Fig. 3, lower panel,  ). It is clear from these ESI data that both the nodE-independent IV(C18:0,Ac) and IV(C18:1,Ac) and the nodEdependent polyunsaturated fatty acid-containing IV(C20:3,Ac) and IV(C20:4,Ac) LCOs are present in our extracts.
Philip-Hollingsworth et al. (8) summarize the results of their mass spectrometric analysis of LCOs in their Table I. Remarkably, ions recorded in negative mode ESI are labeled as M Ϫ ions; e.g. the ion detected at m/z 1094 is assigned as the M Ϫ molecular ion of IV(C18:1,Ac). Unfortunately, this conclusion is based on a critical flaw, since in negative mode ESI, oligosaccharides, and glycoconjugates do not form molecular ions (such as M Ϫ ), but pseudomolecular ions (such as [M Ϫ H] Ϫ ). So, the ion detected at m/z 1094 by the Michigan group, must actually correspond to the [M Ϫ H] Ϫ pseudomolecular ion of IV(C18: 1,Ac), which should give an ion at m/z 1093. The small discrepancy in mass is presumably due to a slight deterioration in the calibration of the mass analyzer.
FAB-MS and CID-MS-MS-Although the [M ϩ H] ϩ pseudomolecular ion of IV(C20:4,Ac) and the [M ϩ Na] ϩ pseudomolecular ion of IV(C18:1,Ac) are isobaric (both having m/z values of 1117), these two species may be very readily distinguished from each other on the basis of their tandem mass spectrometric behavior. In order to demonstrate this, IV(C18:1,Ac) ([M ϩ H] ϩ at m/z 1095) was loaded into a sodium-dosed m-nitrobenzyl alcohol matrix, forcing the LCOs to form sodiated instead of protonated pseudomolecular ions. It should be noted that the nonpooled HPLC fractions described in Ref. 7 were used, in order to be quite certain that only the IV(C18:1,Ac) species was present. In the pseudomolecular ion region of the FAB mass spectrum (Fig. 4B), all [M ϩ H] ϩ pseudomolecular ions are indeed shifted by 22 amu to higher mass, compared with the FAB mass spectrum obtained in the non-sodiated thioglycerol matrix (Fig. 4A), due to the replacement of the proton by a sodium ion. The CID mass spectrum of the [M ϩ H] ϩ ion from IV(C20:4,Ac) (Fig. 5, top panel; already published in Ref. 7, but added here for comparison) shows typical B ions at m/z 490, 693, and 896, well known to be formed from [M ϩ H] ϩ precursor ions (9,15). The CID mass spectrum of the [M ϩ Na] ϩ compound (Fig. 5, bottom panel) is very clearly different and is a classic example of the fragmentation occurring mainly via cross-ring cleavages, characteristic of alkali-cationized oligosaccharide-containing species (12, 16 -18). Most evidently, the suggestion in Ref. 8 that the fragmentation spectrum in Ref. 7 is that of an [M ϩ Na] ϩ pseudomolecular ion can therefore be excluded.
Taken together with the UV absorption data and the previously published mass spectrometric analysis obtained on palladium-catalyzed reduction of the LCOs (7), we feel our findings to be unambiguous proof that R. leguminosarum bv. trifolii does indeed produce nodE-dependent nodulation factors with highly unsaturated fatty acyl moieties.