Rhamnogalacturonan II structure shows variation in the side chains monosaccharide composition and methylation status within and across different plant species


  • Martin Pabst,

    1. Department of Chemistry, University of Natural Resources and Life Sciences, Vienna, Austria
    Current affiliation:
    1. Analytical Chemistry, Laboratory of Organic Chemistry, ETH Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Richard M. Fischl,

    1. Department of Chemistry, University of Natural Resources and Life Sciences, Vienna, Austria
    Current affiliation:
    1. Station Biologique de Roscoff, Roscoff, France
    Search for more papers by this author
  • Lothar Brecker,

    1. Department of Chemistry, University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Willy Morelle,

    1. Unité de Glycobiologie Structurale et Fonctionnelle, Université des Sciences et Technologies de Lille 1, Villeneuve d'Ascq, France
    Search for more papers by this author
  • Alexander Fauland,

    1. Department of Chemistry, 8010 Technical University of Graz, Graz, Austria
    Search for more papers by this author
  • Harald Köfeler,

    1. Core Facility Mass Spectrometry, 8036 Medical University of Graz, Graz, Austria
    Search for more papers by this author
  • Friedrich Altmann,

    1. Department of Chemistry, University of Natural Resources and Life Sciences, Vienna, Austria
    Search for more papers by this author
  • Renaud Léonard

    Corresponding authorCurrent affiliation:
    1. Unité de Glycobiologie Structurale et Fonctionnelle, Université des Sciences et Technologies de Lille 1, Villeneuve d'Ascq, France
    • Department of Chemistry, University of Natural Resources and Life Sciences, Vienna, Austria
    Search for more papers by this author

For correspondence (e-mail renaud.leonard@univ-lille1.fr).


A paradigm regarding rhamnogalacturonans II (RGII) is their strictly conserved structure within a given plant. We developed and employed a fast structural characterization method based on chromatography and mass spectrometry, allowing analysis of RGII side chains from microgram amounts of cell wall. We found that RGII structures are much more diverse than so far described. In chain A of wild-type plants, up to 45% of the l–fucose is substituted by l–galactose, a state that is seemingly uncorrelated with RGII dimerization capacity. This led us to completely reinvestigate RGII structures of the Arabidopsis thaliana fucose-deficient mutant mur1, which provided insights into RGII chain A biosynthesis, and suggested that chain A truncation, rather than l–fucose to l–galactose substitution, is responsible for the mur1 dwarf phenotype. Mass spectrometry data for chain A coupled with NMR analysis revealed a high degree of methyl esterification of its glucuronic acid, providing a plausible explanation for the puzzling RGII antibody recognition. The β–galacturonic acid of chain A exhibits up to two methyl etherifications in an organ-specific manner. Combined with variation in the length of side chain B, this gives rise to a family of RGII structures instead of the unique structure described up to now. These findings pave the way for studies on the physiological roles of modulation of RGII composition.


Rhamnogalacturonan II constitutes approximately 10% of the total pectin amount in the plant cell wall of higher plants (O'Neill et al., 2004). It plays a crucial role in their cell-wall integrity by enhancing cross-linking of the pectic network through formation of RGII dimers binding two RGII monomers via a borate di-ester bound. As the most complex plant polysaccharide, it comprises an α1,4–homogalacturonan backbone of seven to nine residues substituted by up to five side chains designed by the letters A–E and comprising 12 monosaccharides bound by 20 linkages (Whitcombe et al., 1995; Pellerin et al., 1996; Pérez et al., 2003). Chain E comprises one arabinosyl residue and chains C and D are disaccharides. In contrast, the two largest chains A and B in A. thaliana have been described as highly ramified octa- and heptasaccharides, respectively (Hervé du Penhoat et al., 1999; Glushka et al., 2003). Several monosaccharides of these chains are methylated and/or acetylated (Whitcombe et al., 1995). Despite its complexity, the structure of RGII is highly conserved throughout the plant kingdom (Ishii and Kaneko, 1998; Matsunaga et al., 2004). A dramatic decrease in RGII dimer formation, affecting the macromolecular pectin network formation, has been described as a consequence of replacement of one monosaccharide of this structure by another (Reuhs et al., 2004). Alteration of RGII integrity leads to dwarfism (O'Neill et al., 2001), defects in pollen tube elongation (Delmas et al., 2008) and even plant cell death (Ahn et al., 2006).

Despite this structural conservation, some RGII variability has been found across plant species. For instance, the chain red wine is correct. The RGIIs are also present in grapes but the RGII are more easily purified from red wine. B described in A. thaliana is a heptasaccharide, while in red wine, it is a nonasaccharide that also contains one rhamnose and one arabinose residue (Glushka et al., 2003). Deviation from the standard structure also affects lycophytes and pteridophytes, in which 3–O–methyl rhamnosyl residues replace terminal rhamnoses within side chain B (Matsunaga et al., 2004). The existence of a series of three chains A separated by 14 Da, suggesting methyl esterification of the two galacturonic acids of this side chain, has been reported (Séveno et al., 2009), but this modification, which is located in close proximity to the crucial boron-binding apiose residue responsible for RGII dimerization (Ishii et al., 1999), required further characterization.

The original aim of this study was to establish a fast method for analyses of mutants affected in the biosynthesis of RGII. Currently available mass spectrometry methods do not provide separation of the various structures before their analysis and therefore require time-consuming extraction and chromatographic steps starting from large amounts of cell wall material (Melton et al., 1986; Séveno et al., 2009). This leads to difficulties in data interpretation because of superposition of isobaric structures with the same mass but different compositions or arrangements, and because masses generated by in- or post-source fragmentation cannot be distinguished from structures truncated in vivo. Additionally, the need for large amounts of purified pectin for direct infusion experiments is incompatible with fast screening of broad mutant populations.

Using red wine RGII as the standard material, we developed a strategy based on detection by PGCC-ESI-MS (porous graphitic carbon chromatography/electrospray/mass spectrometry) of individual side chains directly released by dilute trifluoroacetic acid (TFA) from insoluble cell-wall material with or without sodium borohydride reduction. The same chromatographic and detection strategy was used for enzymatically released full RGII monomers. These strategies were applied to A. thaliana and to other plants, from mosses to angiosperms, in order to obtain information on the variety of RGIIs. These studies provided evidence that, depending on the plant, between 1 and 45% of the RGII side chains A contain a 3,4-substituted l–galactose residue rather than a 3,4-substituted α–l-fucose. A large portion of the GlcA residues are methyl-esterified, and some of the β–GalA residues are O–methylated at positions 3 and 4 (see Figure 1a and Table 1 for masses and nomenclature). We also provide data showing that the number of rhamnosyl and arabinosyl residues present on side chain B (see Figure 1b and Table 1) varies depending on the age of the plant and the tissue type. When applied to the fucose-deficient mutant mur1, this methodology raises a hypothesis regarding the RGII alteration that leads to the dwarf phenotype, and provides information on the order of assembly of chain A.

Table 1. Nomenclature for the structures described in this study and corresponding observed masses
Structure nameFormulaM+17a nativeM+17a reduced
  1. The acetylation status of B chains is not included in the nomenclature, but most of these chains possess acetylated fucose and aceric acid. Reduction is not used for characterization of B chain structures. NA, not applicable.

  2. a

    Ammonia adduct.

Figure 1.

Nomenclature of side chains A (a) and B (b) used in this study.

The first letter indicates whether the chain is chain A or chain B. For chain A, the first three superscript letters are either Fuc or Gal, depending whether the chain contains an α1,4–fucosyl or an α1,4–galactosyl residue, followed by either MeGlcA or GlcA, depending on whether the glucuronosyl residue is methyl esterified or not. The number from 0–2 indicates the number of methyl etherifications carried by the β–galacturonic acid.


Establishing LC-ESI-MS/MS conditions for separation and identification of RGII side chains

Cleavage of the acid-labile glycosidic bond that links side chains A–D to the polygalacturonic acid backbone is commonly used for their release (Whitcombe et al., 1995; Ishii and Kaneko, 1998; Glushka et al., 2003; Reuhs et al., 2004; Séveno et al., 2009). When using plant cell wall as starting material, and in order to make the sample preparation as easy as possible, we used only two washing steps for the ground plant material, i.e. with 80% ethanol and 100% acetone successively. LC-MS data were obtained efficiently from as little as 500 μg cell-wall material (15 mg fresh plant material).

Séveno et al. (2009) showed that, for ESI-MS fingerprinting without prior capLC (capillary liquid chromatography) separation, it is preferable to perform the hydrolysis in two steps in presence of 0.1 m TFA, first for 16 h at 40°C and then on the remaining insoluble material at 80°C for 1 h. When MS analysis is coupled with LC, such precautions are no longer necessary. No significant difference in the data was observed whether the hydrolysis was performed for 16 h at 40°C or for 1 h at 80°C or whether these two steps were performed successively. For this reason, it is more convenient to restrict the hydrolysis to 1 h at 80°C with 0.1 m TFA. In this study, the hydrolyzed chains were usually reduced by sodium borohydride. This treatment opens the ring of the monosaccharide situated at the reducing end of a chain, avoiding the inconvenience of facing the presence of both α and β isomers. This allows easier interpretation of the MS/MS spectra, and confers higher stability to the chains during preparative purification for NMR analysis. Alternatively, samples were used as for capLC-ESI-MS analysis without prior reduction to preserve the base-labile methyl esterification and acetylation.

Using released chains from red wine RGII as a standard, a single gradient was devised that facilitated separation and identification of the four side chains within 30 min (Figure 2). Chain B exhibited several peaks that potentially result from degradation during the sample preparation or reflect the presence of chains of various lengths in red wine RGII itself. To discriminate between these two possibilities, we performed mild acid hydrolysis for various periods of time ranging from 30 min to 2 h but did not observe changes in the structure ratios greater than a few per cent. We therefore conclude that the diversity observed reflects the variability originally present, and that red wine RGII possesses a panel of chains B ranging from the complete chain B (BRRA) as described by Glushka et al. (2003) to oligosaccharides lacking the distal arabinose residue (BRR) and in some cases even lacking one (BR) or two (BU) rhamnose residues (see Figure 1b and Table 1 for nomenclature).

Figure 2.

BPI (base peak intensity) profile obtained for released side chains A–D corresponding to 1 mg of red wine RGII.

Selected ion traces of the representative structures are shown below. The complete side chain B described for red wine RGII only represents a small minority of the side chains B present in red wine, and many truncated structures are present that lack one or several monosaccharides. Chain A shows substitution of l–fucose by l–galactose, as well as diversity due to the methylation status of one galacturonic acid residue. The nomenclature used is explained in Figure 1 and Table 1. The extracted ion masses were 1326, 1310, 372, 402, 1352, 1220, 1206, 1074 and 928 Da. Peaks corresponding to the extracted masses are indicated by an asterisk. The superscripts 1 and 2 after the asterisks indicate the elution positions of chain B with one additional O–acetylation (aceric acid or O–Me-fucose)or double O–acetylation, respectively. Not all detected chain A structures are shown. Separation of chains A according to their methylation status is shown in more detail in Figure 3.

The α1,4–l-fucose residue of chain A is partly substituted by l–galactose in wild-type plants

When assessing potential variability in the side chain A from red wine RGII, we detected a chain possessing a mass 16 Da higher than expected (Figures 2 and S1). The corresponding structure (here called AGal) elutes earlier and is approximately three times less abundant than the chain A previously described (Figure 2). When subjected to MS/MS analysis by collision-induced dissociation, the parent ion of AGal from red wine (1314.4 Da after borohydride reduction, present as an ammonia adduct [M+17]+ ion) generates major fragments of 475, 651 and 813 Da (Figure S2A). The 651 Da fragment corresponds to the GalA-(GalA)-Rha-Api core, and the 813 Da fragment may be explained by addition of a hexose-like l–galactose (Figure S2A and Table 2). As expected, the structure AFuc from red wine (1298 Da, borohydride-reduced) shows the same 475 and 651 Da fragments, but also a fragment representing addition of a fucose residue (Figure S2C and Table 2). The fragmentation pattern and retention time of the red wine AGal structures are exactly the same as found for mur1 chain A of the same mass, for which it has been previously shown (Reuhs et al., 2004) that l–galactose replaces l–fucose (Figure S2B). From this concomitance of retention time and fragmentation pattern, it may be concluded that the structures are identical and that the wild-type chain A contains l–galactose instead of l–fucose, as described for the mur1 mutant. This is in agreement with the results described by Pellerin et al. (1996), who reported the presence of 3,4-substituted galactose in red wine RGII without identifying its location on a particular chain. In order to ensure that the presence of AGal is not limited to red wine RGII, we screened cell walls from various plants and found that AGal was present in almost all tested samples in relevant amounts, as shown in Figure S3 and Table 3. In the fern Asplenium nidus, AFuc and AGal were present in approximately equal amounts. AGal also represents a high proportion of chain A in tomato (Solanum lycopersicum) fruits and the gymnosperm Cupressus macrocarpa, where it constitutes approximately a quarter of chain A, as in red wine. Replacement of l–fucose by l–galactose occurs in significant amounts in some dicotyledons such as carrot (Daucus carota) and corn salad (Valerianella locusta), in approximately 10–15% of the structures, but is much more scarce in others such as Nicotiana benthamiana (approximately 1–5%). In Brassicaceae, AGal appears to be present only as traces, as indicated by the two plants tested: A. thaliana and cabbage (Brassica oleracea).

Table 2. Series of fragmentation and deduced structures showing replacement of l–fucose by l–galactose in some RGII side chains A from red wine
Mass detected after mild reductionSeries of structures deduced from significant fragmentsDeduced structure
1314.4 math formula AGalGlcA0
1298.5 math formula AFucGlcA0
Table 3. Proportion of chain A (percentage of the total amount) with or without substitution of l–fucose by l–galactose in various plants from moss to higher plants
  1. Unless otherwise mentioned, the studied organs are leaves. Red wine was included as a common source of RGII. The corresponding spectra are presented in Figure S3.

Red wine2476
Solanum lycopersicum 2575
Nicotiana benthamiana 496
Arabidopsis thaliana 0.599.5
Brassica oleracea 0.399.7
Daucus carota 1486
Valerianella locusta 1684
Apium graveolens 4357
Cupressus macrocarpa 2575
Asplenium nidus 4555

One of the two galacturonic acid residues of chain A is O–methylated up to twice

In A. thaliana RGII, for which AGal amounts may be neglected, several peaks separated by 14 Da are visible for chain A, as previously reported (Séveno et al., 2009). The selected ion current chromatograms for the various masses obtained after dilute TFA hydrolysis allowed two series (a and b) of three peaks each to be distinguished, which differed by their retention times (Figure 3, upper panels). The second series (series b in Figure 3) is discussed in the subsequent section.

Figure 3.

Single ion mass obtained for masses corresponding to various chains A from 3-week-old A. thaliana plants.

The upper panels relate to released side chains without further treatment; the lower panels correspond to released side chains treated with reducing agents under mild basic conditions. The mild reducing treatment allows methyl esterification (on carbon 6) to be distinguished from other methylations. Reduction into the neutral sugar only occurs when the uronic acid is methyl esterified, as loss of methanol, together with its oxygen, generates an intermediate that is very sensitive to reducing agents. Reduction of a non-esterified uronic acid to its respective neutral sugar requires much more stringent reducing conditions. The peaks marked ‘o’ do not correspond to RGII structures.

The structures corresponding to the masses of series a (1a, 2a, 3a) showed masses that were increased, as expected, by 2 Da when reduced, due to gain of two hydrogens during opening of the ring of the terminal apiose residue (Figure 3, lower panels, series a). The fragmentation pattern of the structure of mass 1298.5 obtained after sodium borohydride reduction (AFucGlcA0, see Figure 1a and Table 1 for nomenclature) corresponds to the profile expected for the previously described chain A (Figure S4C). Starting from GalA-Rha-Api (475.2 Da), the fragmentation pattern shows addition of a second galacturonic acid residue (Table 4). The chain A with a mass 14 Da higher than the previously described structure (Figure S4B) also generates the fragment GalA-Rha-Api, but the mass of this fragment has increased by 190 Da, corresponding to a mono-O–methyl GalA (Table 4). Confirmation is provided by the presence of a second fragmentation series for this structure AFucGlcA1, starting with a mass of 489.2 Da corresponding to the structure MeGalA-Rha-Api (Table 4). The mass of this ion increases by 176 Da to 665.2 Da, joining the previous series at the structure MeGalA-(GalA)-Rha-Api. The parent ion 1326, corresponding to the structure AFucGlcA2 (Figure S4A) also generates two series of fragment ions (Table 4). As for other chain A structures, the first one starts with a fragment of mass 475.2 Da (GalA-Rha-Api), but continues by addition of 204 Da, corresponding to a GalA residue harboring two methylations, leading to the structure Me2GalA-(GalA)-Rha-Api (mass 679.2 Da). This is confirmed by the second series present in this spectrum, starting at mass 503.2 Da, i.e. 28 Da more than the known chain A. This ion is derived from the fragment Me2GalA-Rha-Api (679.2 Da), the core of the structure AFucGlcA2, by loss of 176 Da (i.e. one GalA). As the two methyl groups present on this GalA residue are not sensitive to reduction under mild basic conditions, it may be deduced that they are linked by an ether bond rather than an ester bond as commonly found on homogalacturonans.

Table 4. Series of fragmentation and deduced structures indicating the presence of up to two methyl etherifications of one galacturonic acid residue of chain A
Peak of selected ion current chromatogramMass after mild reductionStructures deduced from significant fragmentsDeduced structure
1a1326.5math formula and math formulaAFucGlcA2
2a1312.5math formula and math formula AFucGlcA1
3a1298.5 math formula AFucGlcA0

NMR measurements (see Methods S1) performed on purified chain A support the data obtained from mass spectrometric measurements. The 1H-NMR spectrum, as well as COSY, TOCSY and ROESY experiments, despite their poor quality due to the low amounts of material, allow identification of the eight monosaccharides constituting the chain A as well as their inter-glycosidic linkages. These analyses allowed detection of the methyl group bound to α–Xyl in position 2 as previously described (Darvill et al., 1978). Furthermore, the spectra show the presence of two ether methyl groups in slightly sub-stoichiometric amounts, which are both located on the β–GalA (Figure S5). However, we were unable to strictly determine their positions on this glycoside by NMR. To circumvent these difficulties, GC-MS/MS experiments were performed on ICD3 permethylated samples (see Methods S2). The fragments at masses of 164 and 208 Da are compatible with methylation of the β–GalA at positions 3 and 4, whereas the absence of a fragment of mass 118 Da excludes methylation at position 2 of this glycoside (Figure S6).

AFucGlcA1 and AFucGlcA2 are present in all plants that we tested. As the ensemble of chain A modifications is cumulative, replacement of fucose by galactose leads to the structures AGalGlcA1 and AGalGlcA2.

During plantlet development, the level of β–galacturonic acid methylation does not change significantly, affecting between 45 and 50% of the residues. Nevertheless, in adult organs, this rate differs from organ to organ (Figure 4), and is more than twice as high in stems (60%) as in leaves (25%), with siliques showing an intermediate level (45%).

Figure 4.

β–glucuronic acid and β–galacturonic acid methylation rate in organs of adult A. thaliana plants.

With regard to the β–galacturonic acid, most of the methylation occurs at two positions at once. The experiments were replicated twice independently, and each sample was measured twice. Values are means ± SD.

The glucuronic acid residue of chain A is mostly methyl esterified

As previously mentioned, in addition to the fucose-containing series of peaks separated by 14 Da and gaining the classical 2 Da when reduced under mild basic conditions due to opening of the apiose sugar ring, a second series of three peaks with different retention times was observed (upper panels of Figure 3, series b). This series, containing the structures of masses 1310.5, 1324.5 and 1338.5 Da, is referred to as AFucMeGlcA0 to AFucMeGlcA2. When treated with the reducing agent NaBH4 at a slightly increased pH, whereas AFucGlcA0 to AFucGlcA2 increased by 2 Da corresponding to reduction of the apiose residue of these chains, AFucMeGlcA0 to AFucMeGlcA2 decreased by 26 Da (lower panels of Figure 3, series c). Based on the study by Kim and Carpita (1992), and taking in account the 2 Da due to reduction of the apiose residue, this corresponds to loss of methanol and concomitant selective reduction of the methyl esterified uronic acid by NaBH4, leading to the respective neutral sugar (generation of glucose in the case of RGII chain A). This process only occurs when the uronic acid is methyl esterified, as loss of methanol, together with its oxygen, generates an intermediate that is very sensitive to reducing agents. Reduction of a non-esterified uronic acid to its respective neutral sugar requires much more stringent reducing conditions.

When subjected to MS/MS analysis, the structure of mass 1338.5 Da shown in the upper panel of Figure 3 generates series of fragments of which the smallest has a mass of 527.2 Da and corresponds to the structure Rha-(GalA)-Me2GalA (Figure S7 and Table 5). An apiose residue and a fucose residue may be successively added to this fragment to produce Fuc-(Me2GalA)-(GalA)-Rha-Api, but thereafter the mass increment corresponds to addition of methylated glucuronic acid rather than acidic glucuronic acid. This characteristic was found for all methyl esterified structures. The proportion of methyl esterified glucuronic acid varies between 85 and 90% during plant development and does not appear to be organ-specific, as the levels found in leaves, stems and siliques are identical, at approximately 90% (Figure 4).

Table 5. Series of fragmentation and deduced structures indicating the presence of a methyl esterified glucuronic acid residue in some chains A of RGII
Detected mass Series of structures deduced from significant fragmentsDeduced structure
  1. The corresponding MS/MS data are presented in Figure S7.

1338.5 image_n/tpj12271-gra-0001.png AFucMeGlcA2

The length of chain B from A. thaliana RGII varies according to organ and developmental stage

As observed with red wine RGII, variations exist in A. thaliana concerning the length of chain B as well as its degree of acetylation. Chain B with no acetylation or acetylated at one or two positions may be found. Nevertheless, no organ-specific acetylation profile was observed and acetylation affects more than 90% of side chains B in all studied organs. The situation is different for glycosylation, specifically rhamnosylation, whose proportion is specific for the stage of development and the organ.

The main structure found in imbibed seeds is the complete chain B, here referred to as BRR (as A. thaliana lacks BRRA, see Figure 1b and Table 1 for nomenclature). The proportion of this structure, which was previously thought to be absent in A. thaliana (Glushka et al., 2003), decreases from 45% of total chain B to approximately 7% in 18-day-old plants (Figure 5). Single rhamnosylation also decreases but to a much lesser extent, from slightly less than 45% to approximately 33% of the structures. Meanwhile, the proportion of side chains B lacking any rhamnosylation increases considerably from less than 10% to more than 40%. Similar results are obtained whether the plants are grown in the light (Figure 5) or in the dark.

Figure 5.

Variation of chain B terminal rhamnosylation during the growth of A. thaliana plants.

The relative abundance of appropriate ions is shown. Several tens of plants were used for each assay. The experiments were replicated twice independently, and each sample was measured twice. Values are means ± SD.

In adult plants, the rhamnosylation of chain B differs according to the organ considered. The profile is very similar for stems and siliques, the main structure being non-rhamnosylated BU, with other structures clearly under-represented (Figure 6). In adult leaves, the situation is different, as single rhamnosylation is predominant.

Figure 6.

Level of rhamnosylation of RGII chain B in various organs of 6-week-old adult A. thaliana plants.

The experiments were replicated twice independently, and each sample was measured twice. Values are means ± SD.

Concomitant detection of the four side chains of a monomer

The isolation and analysis procedure used for chains A–D provides useful insight into the structural composition of these chains, but provides no information regarding their combination within a given RGII monomer. Therefore, adaptation of the described analytical method to full RGII monomers was undertaken. The basic LC analysis system for side chains was retained and full red wine RGII was used as a standard. Peaks derived from the full structures were further detected in negative mode. Taking into account the already characterized variety of chains A and B, it is not surprising to observe a plethora of combinations of individual side chains, leading to a very complex mass spectrum. The fragment spectra of the peaks from 3500–5000 Da clearly correspond to RGII oligosaccharides (Figure 7a). Several sets of peaks may be observed separated by 176 Da, which originate from the variable length of the oligogalacturonide backbone (seven to nine monosaccharides). Each set of peaks confirms the diversity of structures found in terms of rhamnosylation of chain B, as well as variability of chain A. The MS/MS spectrum of these peaks sheds light on the structures of the four chains at once (Figure 7b). Additionally, it shows that the internal galacturonic acids of the backbone lack methyl esterification.

Figure 7.

LC-ESI-MS and MS/MS analysis of full RGII monomers.

(a) Summarized mass spectrum of triply charged ions obtained in negative mode after Rapidase® digestion of 2 mg of A. thaliana cell wall. The variety of masses derives from heterogeneity of the GalA backbone and its terminal methylation as well as of chains A and B. The peak clusters originate from methylation of β–GalA and esterification of GlcA residues.

(b) The MS/MS spectrum of mass 1566 (M-3H)3- in negative ion detection mode. All four side chains were detected after collision-induced dissociation as singly negatively charged masses (chain D, 335 Da; chain C, 365 Da, BR twice acetylated, 1121 Da; fucose containing chain A with one methyl on one uronic acid, 1273 Da).

Neither the calculation of masses of the many theoretically possible combinations nor MS/MS fragmentation analysis revealed the presence of chain E, possibly because of its hydrolysis during Rapidase® digestion.

Application to the A. thaliana mutant mur1

One of the main findings of this study is that substitution of l–fucose by l–galactose naturally occurs in wild-type plants. In tomato plants, it represents approximately a fifth of the structures; however, the dimerization rate of RGII was approximately 97% (Voxeur et al., 2011). This result is incompatible with the possibility that substitution of l–fucose by l–galactose is responsible for the defect in RGII dimerization (and therefore the dwarfism) in the fucose-deficient mutant mur1. For this reason, we decided to reinvestigate the side chains A present in this mutant. As described previously in mur1 (Reuhs et al., 2004), a fragment with a mass corresponding to substitution of l–fucose by l–galactose is present at 1314.6 Da, whereas the structure containing l–fucose is not detectable in aerial organs (Table 6 and Figure S8). Nevertheless, other masses of more or less truncated chains A populate the spectrum. The highly truncated structure, exclusively composed of the apiose residue, one rhamnose and two galacturonic acids, is actually the most dominant chain A in this plant (Table 6 and Figure S8). Our hypothesis is that the defect of RGII dimerization in mur1 does not result from substitution of l–fucose by l–galactose but instead from the predominance of very truncated RGII structures in this plant. Intermediate structures are normally not present in wild-type plants, and their occurrence in mur1 potentially derives from the lower affinity towards l–galactose donor or acceptor substrates of enzymes responsible for RGII biosynthesis in A. thaliana. The presence of GDP-l–galactose as well as nascent chain A containing l–galactose instead of l–fucose may both slow down elongation of chain A, at least in this plant. In mur1, it is therefore possible to establish in which order some monosaccharides are added to the nascent chain A. Only structures that have been xylosylated may be further extended by a glucuronic acid. In mur1, the xylose added to the nascent chain A is mostly non-methylated, as emphasized by the presence of the mass at 962 Da corresponding to a truncated chain A lacking both the terminal galactose and the glucuronic acid, in which no sugar is methylated, not even the xylose.

Table 6. Comparison of the chain A structures detected in A. thaliana wild-type and mur1 mutant, highlighting the predominance of strongly truncated chains A in mur1
StructureMassPresence in A. thaliana wild-typePresence in A. thaliana mur1 mutant
  1. The corresponding MS data are presented in Figure S8.

GalA-(MeGalA)-Rha-Api +++
Gal-(GalA)-(MeGalA)-Rha-Api +
Xyl-Gal-(GalA)-(MeGalA)-Rha-Api +
GlcA-(Xyl)-Gal-(GalA)-(MeGalA)-Rha-Api +


RGII structures have been elucidated by NMR and mass spectrometry performed on large amounts of highly purified pectin, and methods allowing routine analysis of a few hundred milligrams of RGII side chains by direct-infusion MS/MS fingerprinting have been described (Séveno et al., 2009). However, these methods involve chromatographic steps preceding the analysis, and are therefore not optimal for dealing with a high number of samples in the case of mutant screening. Furthermore, direct-infusion methods do not allow discrimination between isobaric structures that differ in the position of methyl or acetyl groups or even sugar residues (galacturonic acids in α1,2 or β1,3 on chain A; rhamnose in α1,2 or α1,3 on chain B). Finally, fingerprinting methods available up to now were not suitable for quantitative analysis. Our approach eliminates these inconveniences. A key advantage of our method is that structures are separated by LC before detection by MS, which ensures that the observed truncated structures are not generated by fragmentation during analysis but are in fact present in the analyzed sample. This aspect was particularly crucial in the case of re-investigation of the fucose-deficient mutant mur1 in which chain A is predominantly present as a tetrasaccharide, i.e. the apiose residue elongated by a rhamnose and two galacturonic acids. It was thought that the dwarfism of the aerial part of mur1 was the result of a defect in RGII biosynthesis manifested by replacement of the α1,4-linked l–fucose of chain A by l–galactose (Reuhs et al., 2004) leading to a decreased affinity for boron. Instead, our experiments indicate that this defect of dimerization simply results from the presence of truncated chains. As intermediary structures have been identified between the tetrasaccharide and the full chain A, it may be interesting to purify monomeric and dimeric RGII from this mutant and to characterize more precisely the chains A that are present within these two populations. Recently, Voxeur et al. (2011) characterized a tomato plant whose gene for GDP-d–mannose 3,5–epimerase was down-regulated. As for mur1, the plant growth was affected, but this defect was additionally accompanied by fragility of the stems and fruits (Gilbert et al., 2009). Analysis of the cell-wall composition of this knockdown plant revealed the presence of RGII whose side chain A lacks terminal l–galactose. The authors attributed the observed phenotype to a defect in RGII dimerization caused by the absence of terminal l–galactose. Our results provide different interpretation for these phenotypes by indicating the presence of l–galactose in a more internal position of side chain A for approximately 25% of the structure. The truncation of side chain A due to absence of this internal l–galactose is likely to be responsible for the observed phenotypes. As for mur1, precise composition analysis of monomeric and dimeric RGII may resolve this ambiguity.

A monoclonal antibody (CCRC–R1) has been generated against RGII by phage display (Williams et al., 1996), and a polyclonal antibody recognizing RGII has been produced in rabbits (Matoh et al., 1998). Both antibodies were much more efficient after base treatment of the samples, allowing RGII de-esterification. As it was not known that the glucuronic acid of chain A is mostly methyl esterified, no clear explanation was given for this phenomenon, and it was suggested that the methyl esterified polygalacturonic acid backbone was part of the recognized epitope (Willats et al., 2001), even though de-esterified homogalacturonan in itself is not recognized by CCRC-R1. The full monomer analysis performed here by capLC-ESI-MS/MS showed methyl esterification only on galacturonic acids that were two to three residues away from the outer side chains, but did not suggest methyl esterification at the proximate or internal galacturonic acid residues. We suggest that de-esterified chain A may be the real epitope, and that methyl esterification of its glucuronic acid may be the reason for the decreased binding affinity. As Williams et al. (1996) found that plasma membrane-proximal RGII lacked the esters masking CCRC–R1 binding, whereas these esters were present in RGII in the rest of the cell wall, this may reflect distribution of glucuronic acid methyl esterification of side chain A within the plant cell wall. Inhibition studies with purified methyl esterified or acidic side chains A may help to confirm or invalidate this hypothesis.

In most cases, the study of RGII biosynthesis is complicated by the impossibility of performing glycosyltransferase tests due to absence of the proper enzymatic substrate. Chemical synthesis of some substrates has been performed (Rao et al., 2008; Nepogodiev et al., 2011), but the great complexity of RGII makes performing such synthesis for each potential reaction an insurmountable challenge. An alternative involving use of a monosaccharide as an acceptor substrate has been used successfully for chain A-specific xylosyltransferase (Egelund et al., 2006), but the presence of chains A truncated at various levels in the mur1 mutant as well as the existence of chains B of variable length in wild-type plants may make it possible to obtain specific substrates more easily through purification by HPLC using graphitized carbon column as has been performed here for the methylated side chains A.

Up to now, RGII were considered to possess a unique structure within a given plant species, leaving no place for regulation of any physiological aspect by modulation of the RGII composition. This study shows that variation of RGII within a single individual plant does occur and involves monosaccharide substitution and methylation of chains A and variation in the length of chains B. This study opens the door to studies into the physiological roles of modulation of RGII composition.

Experimental procedures

Plant material

Except for A. thaliana, the plant material used in this study was collected either in the Botanical Garden of Vienna or obtained from shops. Seeds from A. thaliana Columbia (Col–0), kindly supplied by Richard Strasser (Department of Applied Genetics und Cell Biology, University of Natural Resources and Life Sciences, Vienna, Austria) and mur1 (NASC ID N6244, European Arabidopsis Stock Center, http://arabidopsis.info/) were surface-sterilized using 0.15% Tween–20 and 0.072% sodium hypochlorite for 15 min on a rotary plate. The seeds were washed four times with sterile water to remove sodium hypochlorite, and plated on sterile soil comprising perlite and common potting soil. The seeds were stratified at 4°C for 2 days, and then incubated in a common plant growth chamber at 22°C constant temperature, 50% humidity and a 16 h light period (long day).

The mur1 mutant was screened for homozygous knockout plants by PCR after extraction of genomic DNA using a Plant Genomic DNA Miniprep Kit (Sigma-Aldrich, http://www.sigmaaldrich.com).

Plant cell-wall preparation

Red wine RGII was kindly supplied by Malcolm A. O'Neill (Complex Carbohydrate Research Center, The University of Georgia, Athens, GA). Plant material (typically 300 mg fresh weight) was frozen in liquid nitrogen before grinding in an Eppendorf tube using a Retsch milling machine (model MM200; Retsch, http://www.retsch.com) at 25 Hz in the presence of two iron bullets for 2 min. For preparation from larger amounts, the material was ground in liquid nitrogen with a mortar and pestle.

The plant cell-wall preparation was performed as described by Matsunaga et al. (2004) with modifications. The powder was homogenized in 80% ethanol before centrifugation for 5 min at 10 000 g at room temperature. The supernatant was removed, and the extraction was repeated once with 80% ethanol and twice with pure acetone. The remaining material was dried in a speed vac (leading typically to 10 mg dry weight). A flow diagram summarizing the steps of the plant cell-wall preparation and the release of RGII side chains or monomers is presented in Figure S9.

Release and sodium borohydride reduction of RGII side chains or monomer

Rhamnogalacturonan chains were liberated by mild acid hydrolysis at 80°C for 1 h in the presence of 0.1 m TFA (100 μl TFA per mg dry ethanol/acetone-insoluble cell-wall residue) either in 2 ml Eppendorf tubes or in 15 ml Corning tubes depending on the amount used. Enzymatic release of RGII was performed overnight using Rapidase® (Rapidase® Vegetable Juice, DSM Food Specialties, http://www.dsm.com) at 37°C and pH 5 in McIlvaine buffer (0.2 m Na2HPO4, 0.1 m citric acid, pH 6.0).

Reduction of RGII side chains was achieved after basification of the samples using ammonium and incubation for 2 h at room temperature in 500 μl of a 0.5% sodium borohydride solution.

Mild base treatment of purified chain A was performed in presence of 0.05–0.1 m NaOH at room temperature for 15 min.

Pre-purification of the sample prior to LC-ESI-MS/MS analysis was performed using an SPE-PGC cartridge (Hypercarb 10 mg column, Thermo-Fisher Scientific, http://www.thermofisher.com) as described previously (Packer et al., 1998; Pabst and Altmann, 2008; Pabst et al., 2010). Before injection to the LC-MS analysis system, the purified samples were dried by speed vac and dissolved in 0.3% formic acid pH 3.


Liquid chromatography was performed on porous graphitized carbon as the stationary phase (0.32 × 150 mm, Hypercarb, Thermo-Fisher Scientific). The elution gradient was set from 100% aqueous buffer (ammonium formiate buffer, 0.3% formic acid buffered with ammonia to pH 3) to 30% acetonitril within 30 min, employing a capillary chromatography system from Dionex (http://www.dionex.com, NCP, Ultimate 3000).

Mass detection was performed by online coupling via an electrospray ion source to a Q–TOF mass spectrometer (Q–TOF Global Ultima, Waters, www.waters.com). Detection of released chains was performed in ES-positive mode, whereas total monomers were analyzed in ES-negative mode.

Fragmentation analysis was performed using collision-induced dissociation with a collision energy of 30% for positive mode detection and 70% for negative ion detection.

Preparative purification of side chains by HPLC

NMR samples were prepared from approximately 5 g cell wall obtained from Nicotiana benthamiana leaves. TFA hydrolysis for chain A release was performed as described above. The lysate was ultra-centrifuged at 100 000 g for 1 h at room temperature and pre-purified using 250 mg Hypercarb SPE cartridges. Crude cell-wall lysate was further separated by size-exclusion chromatography using a Sephadex G15 column (0.5 × 30 cm [diameter × length]), and chain A-containing fractions were further purified by weak anionic exchange chromatography by HPLC using diethylaminoethyl filled column (TOSHO Biosciences,http://www.tosohbioscience.com). Chain B eluted first, with just one acidic residue, whereas chain A with methyl esterified GlcA eluted slightly before chain A containing free GlcA.

A final purification and desalting step was performed by PGC HPLC, by which it was even possible to separate chain A structures according to the number of O–methylations of the galacturonic acid residue.


We thank Paul Kosma for advice on organic chemistry, Richard Strasser for supplying plant material, and Malcolm A. O'Neill (315 riverbend Road, Athens GA, 30602-4712, USA) from the Complex Carbohydrate Research Center for supplying red wine RGII. We also thank DSM Food Specialties for providing the Rapidase® used in this study. This work was funded by the Austrian Science Fund via project grant P20132-B16 to R.L.