Mixed-linkage (1→3,1→4)-β-d-glucan is a major hemicellulose of Equisetum (horsetail) cell walls


  • Stephen C. Fry,

    1. The Edinburgh Cell Wall Group, Institute of Molecular Plant Sciences, School of Biological Sciences, The University of Edinburgh, Daniel Rutherford Building, The King's Buildings, Edinburgh EH9 3JH, UK
    Search for more papers by this author
  • Bertram H. W. A. Nesselrode,

    1. The Edinburgh Cell Wall Group, Institute of Molecular Plant Sciences, School of Biological Sciences, The University of Edinburgh, Daniel Rutherford Building, The King's Buildings, Edinburgh EH9 3JH, UK
    Search for more papers by this author
  • Janice G. Miller,

    1. The Edinburgh Cell Wall Group, Institute of Molecular Plant Sciences, School of Biological Sciences, The University of Edinburgh, Daniel Rutherford Building, The King's Buildings, Edinburgh EH9 3JH, UK
    Search for more papers by this author
  • Ben R. Mewburn

    1. The Edinburgh Cell Wall Group, Institute of Molecular Plant Sciences, School of Biological Sciences, The University of Edinburgh, Daniel Rutherford Building, The King's Buildings, Edinburgh EH9 3JH, UK
    Search for more papers by this author

Author for correspondence:
Stephen C. Fry
Tel: +44 131 650 5320
Fax: +44 131 650 5392
Email: S.Fry@ed.ac.uk


  • • Mixed-linkage (1→3,1→4)-β-d-glucan (MLG) is a hemicellulose reputedly confined to certain Poales. Here, the taxonomic distribution of MLG, and xyloglucan, especially in early-diverging pteridophytes, has been re-investigated.
  • • Polysaccharides were digested with lichenase and xyloglucan endoglucanase (XEG), which specifically hydrolyse MLG and xyloglucan, respectively. The oligosaccharides produced were analysed by thin-layer chromatography (TLC), high-pressure liquid chromatography (HPLC) and alkaline peeling.
  • • Lichenase yielded oligo-β-glucans from all Equisetum species tested (Equisetum arvense, Equisetum fluviatile, Equisetum scirpoides, Equisetum sylvaticum and Equisetum ×trachyodon). The major product was the tetrasaccharide β-glucosyl-(1→4)-β-glucosyl-(1→4)-β-glucosyl-(1→3)-glucose (G4G4G3G), which was converted to cellotriose by alkali, confirming its structure. Minor products included G3G, G4G3G and a nonasaccharide. By contrast, poalean MLGs yielded G4G3G > G4G4G3G > nonasaccharide > dodecasaccharide. No other pteridophytes tested contained MLG, including Psilotum and eusporangiate ferns. No MLG was found in lycopodiophytes, bryophytes, Chara or Nitella. XEG digestion showed that Equisetum xyloglucan has unusual repeat units.
  • • Equisetum, an exceedingly isolated genus whose closest living relatives diverged > 380 million years ago, has evolved MLG independently of the Poales. Equisetum and poalean MLGs share basic structural motifs but also exhibit clear-cut differences. Equisetum MLG is firmly wall-bound, and may tether neighbouring microfibrils. It is also suggested that MLG acts as a template for silica deposition, characteristic of grasses and horsetails.


The cell walls of land plants (embryophytes) are based on a scaffolding of cellulosic microfibrils interspersed by a matrix of diverse polysaccharides (pectins and hemicelluloses), glycoproteins and sometimes also other substances such as lignin, suberin, cutin, cutan and silica (Fry, 2001). The hemicelluloses are thought to play an important role in this structure by hydrogen-bonding to, and tethering, adjacent microfibrils (Fry, 1989; Hayashi, 1989).

In the primary cell walls of most land plants, the major hemicellulose is xyloglucan. Xyloglucan has been detected in the walls of all land plants tested, including bryophytes and pteridophytes, but it does not appear to occur in charophytic algae (the closest living relatives of the land plants) (Popper & Fry, 2003, 2004). Xyloglucan is a (1→4)-β-d-glucan, many of whose d-glucose (Glc) residues carry an α-d-xylose residue on position 6; some of the xylose residues carry additional sugars, especially d-galactose and l-fucose (Obel et al., 2006). In many plants, the primary structure of the xyloglucan is built up of cellotetraose (G4G4G4G)-based repeat units, three of the four Glc residues carrying a xylose side-chain. These repeat units are released in the form of xyloglucan oligosaccharides upon digestion with cellulase or xyloglucan-specific endoglucanase (XEG) (Pauly et al., 1999).

Other taxonomically widespread hemicelluloses of the primary cell wall are those based on backbones of (1→4)-β-d-xylan, (1→4)-β-d-mannan, glucomannan and glucuronomannan (Brett & Waldron, 1996; Obel et al., 2006). The Poaceae (grasses and cereals) and some related families of the order Poales are generally considered unique among vascular plants in that they contain an additional hemicellulose known as mixed-linkage (1→3),(1→4)-β-d-glucan (MLG) (Labavitch & Ray, 1978; Stone & Clarke, 1992; Popper & Fry, 2004; Trethewey et al., 2005). A survey of land plants failed to reveal MLG in any nonpoalean vascular plants (Popper & Fry, 2004), although a related lichenase-digestible polysaccharide, which contained arabinose as well as Glc residues, occurs in the leafy liverwort Lophocolea bidentata (Popper & Fry, 2003). An MLG, known as lichenan, also occurs in some lichens, for example Cetraria islandica (Stone & Clarke, 1992).

Poalean MLG consists of an unbranched chain of β-d-glucopyranose residues linked by (1→3) and (1→4) bonds. The polysaccharide typically takes the form:

... G3G4G4G3G4G4G4G3G4G4G3G4G4G3G ...

where ‘G’ is β-d-glucopyranose, and ‘3’ and ‘4’ indicate (1→3) and (1→4) bonds, respectively; the underlined domains are effectively cellotriose and (fewer) cellotetraose units interlinked by (1→3) bonds (Meikle et al., 1994).

MLG is readily recognized by its susceptibility to digestion by lichenase ((1→3),(1→4)-β-d-glucan 4-glucanohydrolase; EC, an MLG-specific endoglucanase, which cleaves the (1→4) bond in the sequence ... Glc-(1→3)-Glc-(1→4)-Glc ... , yielding diagnostic oligosaccharides (Meikle et al., 1994; Grishutin et al., 2006; Li et al., 2006). The major oligosaccharide released by lichenase digestion of poalean MLG is the trisaccharide G4G3G; smaller amounts of the tetrasaccharide G4G4G3G are also obtained. A few longer runs of up to 11 consecutive (1→4) bonds have also been reported (Woodward et al., 1983). The (1→3) bonds confer flexibility and water-solubility on MLG (Buliga et al., 1986), whereas a uniformly (1→4)-linked chain (= cellulose) is rigid and water-insoluble. A role for MLG in tethering microfibrils has been suggested (Labavitch & Ray, 1978; Wada & Ray, 1978; Fry, 1989).

We recently observed that the early-diverging pteridophyte genus Equisetum possesses a novel endotransglucosylase activity that is similar to xyloglucan endotransglucosylase (XET; Thompson & Fry, 2001) but that uses MLG as its donor substrate and xyloglucan as its acceptor substrate (Fry et al., in press). This discovery led us to re-investigate Equisetum and related plants for the possible occurrence of MLG, and to check the occurrence of xyloglucan in them.

Equisetum is the sole surviving genus of the Equisetopsida (Bell & Hemsley, 2000), a formerly diverse and dominant class of pteridophytes. There are c. 15 extant Equisetum species and several hybrids (Des Marais et al., 2003). The earliest fossil Equisetopsida are from the upper Devonian (377–363 million years ago (Ma)); they were abundant in the Carboniferous, with some (e.g. Calamites) forming forests of 20-m trees. They are eusporangiate euphyllophytes, generally regarded as possessing small megaphylls as opposed to the microphylls characteristic of the earlier-diverging Lycopodiopsida. The Equisetopsida is one of the four extant classes (Table 1) of a monophyletic group called the Moniliformopses (‘ferns’, in the broad sense; Pryer et al., 2001); thus horsetails are early-diverging ‘ferns’. Although Equisetopsida is not the earliest-diverging class of the Moniliformopses (that status is thought to be held by the Psilopsida, which includes Psilotum and Ophioglossum; Smith et al., 2006), nevertheless the ancestors of Equisetum diverged from the ancestors of all other living plants at least 370 Ma (Bell & Hemsley, 2000).

Table 1.  Plants surveyed for mixed-linkage (1→3,1→4)-β-d-glucan (MLG)
  • Cell-wall-rich material (AIR) from each specimen was digested with lichenase and the products were analysed by high-pressure liquid chromatography (HPLC) and/or thin-layer chromatography (TLC). The presence of the MLG-diagnostic trisaccharide (G4G3G) and tetrasaccharide (G4G4G3G) is reported on a subjective scale (++++, abundant; +, detectable; –, undetectable).

  • a

    The plants here termed pteridophytes, also known as the Moniliformopses, constitute a monophyletic group (Pryer et al., 2001; Smith et al., 2006).

  • b

    Polypodiopsida are the leptosporangiate ferns; the other pteridophytes listed are eusporangiate.

Chara vulgarisWhole shootsAlga, Charales
Nitella opacaWhole shootsAlga, Charales
Anthoceros agrestisCell cultureBryophyte, hornwort
Pellia epiphyllaThallusBryophyte, liverwort
Pellia epiphyllaSetaBryophyte, liverwort
Pellia epiphyllaCapsuleBryophyte, liverwort
Pleurozia purpureaWhole shootsBryophyte, liverwort
Sphagnum rubellumWhole shootsBryophyte, moss
Drepanocladus sp.Whole shootsBryophyte, moss
Hygrohypnum sp.Whole shootsBryophyte, moss
Hylocomium splendensWhole shootsBryophyte, moss
Mnium hornumWhole shootsBryophyte, moss
Plagiothecium sp.Whole shootsBryophyte, moss
Rhytidiadelphus loreusWhole shootsBryophyte, moss
Diphasiastrum alpinumWhole shootsLycopodiophyte
Huperzia selagoWhole shootsLycopodiophyte
Lycopodium clavatumWhole shootsLycopodiophyte
Lycopodium phlegmariaWhole shootsLycopodiophyte
Selaginella apodaWhole shootsLycopodiophyte
Selaginella willdenoviiWhole shootsLycopodiophyte
Psilotum nudumWhole shootsPteridophyte,a Psilotopsida
Equisetum arvenseWhole shootsPteridophyte, Equisetopsida++++++
Equisetum fluviatileWhole shootsPteridophyte, Equisetopsida++++++
Equisetum scirpoidesWhole shootsPteridophyte, Equisetopsida++++
Equisetum sylvaticumWhole shootsPteridophyte, Equisetopsida++++
Equisetum× trachyodonWhole shootsPteridophyte, Equisetopsida++++
Angiopteris erectaLeavesPteridophyte, Marattiopsida
Osmunda regalisLeavesPteridophyte, Polypodiopsidab
Asplenium trichomanesLeavesPteridophyte, Polypodiopsida
Athyrium filix-feminaLeavesPteridophyte, Polypodiopsida
Polypodium vulgareLeavesPteridophyte, Polypodiopsida
Polystichum aculeatumWhole shootsPteridophyte, Polypodiopsida
Pteridium aquilinumLeavesPteridophyte, Polypodiopsida
Cycas revolutaLeavesGymnosperm
Larix deciduaLeavesGymnosperm
Iris pseudacorusLeavesAngiosperm, nonpoalean monocot
Avena sativaColeoptilesAngiosperm, Poales+++++++
Festuca arundinaceaCell cultureAngiosperm, Poales++++++
Holcus lanatusLeavesAngiosperm, Poales++++++
Hordeum vulgareColeoptilesAngiosperm, Poales++++++
Lolium perenneLeavesAngiosperm, Poales++++++
Zea maysCell cultureAngiosperm, Poales+++++
Rosa sp.Cell cultureAngiosperm, eudicot
Spinacia oleraceaCell cultureAngiosperm, eudicot

Equisetum is anatomically unique among extant plants (Golub & Wetmore, 1948). The stem consists of brittle, readily disconnectable internodes, which (as in the Poales) grow at the extreme base and are rich in silica. At each node is a whorl of tiny leaves. Lateral shoots originate between the leaves of a whorl and not, as in all other plants, in the axils. All roots of Equisetum are adventitious (except the single primary root growing from the zygote), another feature commonly observed in the Poales.

Despite their unique anatomical features, horsetails were not thought to be outside the normal range of cell wall composition for euphyllophytes (Popper & Fry, 2004), although they are relatively rich in galacturonic acid and poor in xylose (Nothnagel & Nothnagel, 2007). However, we now show that this exceedingly isolated genus of plants possesses the allegedly poalean-specific polysaccharide MLG. The evolution of the ability to synthesize this polysaccharide, and its possible biological roles, are discussed.

Materials and Methods

Polysaccharides and oligosaccharides

Tamarind (Tamarindus indica L.) seed xyloglucan was a generous gift from Dr K. Yamatoya, Dainippon Pharmaceutical Co., Osaka, Japan; barley (Hordeum vulgare L.) MLG was from Sigma Chemical Co. Specific cello-oligosaccharides and laminari-oligosaccharides were from Sigma Chemical Co. or Megazyme (Bray, Ireland). Water-soluble cellulose acetate was synthesized by a modification of the method of Gomez-Bujedo et al. (2004). The cello-oligosaccharide marker-mixture used for high-pressure liquid chromatography (HPLC) was prepared by cellulase digestion of water-soluble cellulose acetate followed by de-acetylation in 0.1 M NaOH at room temperature for 1 h and neutralization with acetic acid.

Plant materials

Cell-suspension cultures of spinach (Spinacia oleracea L.), maize (Zea mays L.), tall fescue grass (Festuca arundinacea Schreber) and rose (Rosa sp., cv. Paul's Scarlet) were as used previously in this laboratory (Wende & Fry, 1997; Lindsay & Fry, 2008); those of Anthoceros agrestis Paton (Vogelsang et al., 2006) were a generous gift of Dr Maike Petersen, University of Marburg, Marburg, Germany. Plants of Marchantia polymorpha L., Osmunda regalis L., Polypodium vulgare L., Lolium perenne L., Holcus lanatus L., Equisetum fluviatile L. and Equisetum arvense L. were grown in a private garden in Edinburgh, UK. Iris pseudacorus L., Chara vulgaris L. and Nitella opaca (Bruz.) Agardh were grown in a garden pond in Edinburgh. Cycas revoluta Thunb., Selaginella apoda (L.) Spring and Asplenium trichomanes L. were glasshouse-grown at the King's Buildings, Edinburgh. Equisetum hyemale L., Equisetum scirpoides Michx., Equisetum variegatum Weber & Mohr, Equisetum× trachyodon A.Br., Angiopteris erecta Hoffm., Lycopodium phlegmaria L. and Selaginella willdenovii (Desv. ex Poir.) Baker were generous gifts of the Royal Botanic Garden, Edinburgh. The following plants were wild-collected: Equisetum sylvaticum L., Pellia epiphylla L. (Corda), Drepanocladus sp., Hygrohypnum sp., Hylocomium splendens (Hedw.) Schimp., Rhytidiadelphus loreus (Hedw.) Warnst., Athyrium filix-femina (L.) Roth, Pteridium aquilinum (L.) Kuhn and Larix decidua Miller (collected in Midlothian), Lycopodium clavatum L. (collected in Tweeddale), and Sphagnum rubellum Wils. and Pleurozia purpurea Lindb. (collected in Argyll). All other plants were sourced as described by Popper & Fry (2003, 2004).

Alcohol-insoluble residue (AIR) and hemicellulose B

Techniques were basically as described previously (Fry, 2000). AIR was prepared by thorough homogenization of plant tissue in 75% ethanol and the insoluble matter was washed rigorously with 75% ethanol, then acetone, and then dried. Total hemicellulose was extracted from AIR with 6 M NaOH containing 1% NaBH4 at 37°C for 24 h (Edelmann & Fry, 1992), neutralized with acetic acid, and dialysed against water. The contents of the dialysis sac were centrifuged at 3000 g for 5 min, pelleting the insoluble hemicellulose A; the supernatant (hemicellulose B) was freeze-dried.

Differential extraction of polysaccharides

AIR (100 mg) was shaken for 3 d in 14 ml of dimethylsulphoxide (DMSO) at 20°C, and then bench-centrifuged. The pellet was washed once with another 14 ml of DMSO, and the pooled DMSO solutions were dialysed against water. DMSO-inextractable material was suspended in 45 ml of water and incubated at 100°C for 30 min, and the new extract was dialysed. Both dialysates and the remaining nonextracted material were freeze-dried.

Digestion with lichenase, XEG and Driselase

Lichenase (Megazyme) was dissolved at 7 U ml−1 in a volatile buffer (pyridine/acetic acid/water 1 : 1 : 98, pH 4.7, containing 0.5% chlorobutanol) and added to 10 volumes of a solution or suspension of the sample (typically 0.5% w/v in water). Digestion was stopped after 1–72 h by the addition of 0.1 volumes of 90% formic acid. Products were analysed by thin-layer chromatography (TLC) or HPLC.

Hemicellulose B (5 mg) was treated with 1 ml 0.01% (w/v) XEG (Novo Nordisk A/S, Bagsværd, Denmark) (Pauly et al., 1999) in the above buffer at 37°C for 16 h, and then boiled for 20 min. After drying, a portion was analysed by HPLC. A further 1 mg of hemicellulose B was treated with 200 µl of 0.5% (w/v) Driselase (Sigma Chemical Co.; partially purified as described by Fry (2000)) in the same buffer at 37°C for 3 d, and the digestion was stopped by the addition of 30 µl of 90% formic acid. After drying, a portion was analysed by HPLC.

Alkaline ‘peeling’ of oligosaccharides

Oligosaccharides were incubated in a tightly capped tube with 50 mM NaOH at 20°C for 0–64 h (routinely 4 h), and then neutralized with acetic acid and analysed by TLC.

Thin-layer chromatography

Oligosaccharides produced by lichenase digestion were separated on Merck silica-gel ‘60’ TLC plates developed (usually twice) in butan-1-ol/acetic acid/water (2 : 1 : 1 by volume). The plates were dipped through freshly prepared stain (0.5% (w/v) thymol and 5% (v/v) concentrated H2SO4 in 96% (v/v) ethanol), dried, and heated in an oven at 105°C for 3–6 min.

High-pressure liquid chromatography

Lichenase digests were also analysed by HPLC on a CarboPac PA1 column (‘high-performance’ anion-exchange chromatography, HPAEC; Dionex, Camberley, UK) eluted at 1 ml min−1 with 0.1 M NaOH which also contained: 0–30 min, 0→0.3 M sodium acetate (linear gradient); 30–36 min, 1 M sodium acetate; 36–42 min, 0 M sodium acetate. A pulsed amperometric detector with a gold electrode was used.

XEG digests were analysed on the same apparatus with elution in (1 ml min−1): 0–20 min, 100 mM NaOH/0 mM sodium acetate → 100 mM NaOH/100 mM sodium acetate (linear gradient); 20–25 min, 800 mM NaOH; 25–30 min, 100 mM NaOH.

Driselase digests were analysed on the same apparatus with elution in (1 ml min−1): 0–2 min, 20 mM NaOH; 2–40 min, water; 40–75 min, water → 800 mM NaOH (linear gradient); 75–82 min, 800 mM NaOH; 82–90 min, 20 mM NaOH.


TLC evidence for MLG in Equisetum spp.

Unexpectedly, MLG was readily detectable in horsetails. For example, digestion of E. fluviatile AIR with lichenase yielded several oligosaccharides identical to those from the grass F. arundinacea (Fig. 1).

Figure 1.

Thin-layer chromatography evidence for mixed-linkage (1→3,1→4)-β-d-glucan (MLG) in Equisetum. Polysaccharides from cell cultures of (a) Spinacia oleracea and (b) Festuca arundinacea and from (c) Equisetum fluviatile stems were digested with lichenase for 1 or 16 h. Digests were chromatographed either directly or after NaOH peeling. ‘Enzyme’ tracks received lichenase but no polysaccharides. The right-hand three tracks on each plate show markers: laminari-oligosaccharides (Lam), cello-oligosaccharides (Cell) and xyloglucan-oligosaccharides. Presumed degrees of polymerization of lichenase products are shown (left); 2 and 2′ are an unidentified disaccharide and laminaribiose, respectively. Arrows indicate products formed from these oligosaccharides by alkaline ‘peeling’ for 4 h.

Lichenase cleaves MLG at the (1→4) bond in the sequence ---G3G4G--- (Meikle et al., 1994). Thus the digestion products are all predicted to have a (1→3) bond next to the reducing terminus. In the Poaceae (e.g. Festuca (Fig. 1b) and all other grasses and cereals tested (Table 1)), the major product of lichenase digestion was the trisaccharide G4G3G, followed by the tetrasaccharide G4G4G3G, as expected from known poalean MLG structures. The disaccharide laminaribiose (G3G) was very weakly detectable in some poalean samples. No free glucose was detected, demonstrating the absence of β-glucosidase in the lichenase preparation. Short-term digestion (for 1 h) yielded some intermediate products that largely disappeared after 16 h of digestion – for example, a hexasaccharide (labelled ‘6’ in Fig. 1b), presumed to be G4G3G4G4G3G, which would ultimately be cleaved to produce two molecules of G4G3G (Meikle et al., 1994).

Uniquely among the nonpoalean plants surveyed, Equisetum AIR also gave several of the same oligosaccharides (Fig. 1c). These oligosaccharides were not detectably generated from cultured cells of the dicotyledon S. oleracea (Fig. 1a), or from various other dicotyledons, nonpoalean monocots, or numerous nonangiosperms including gymnosperms, pteridophytes, bryophytes and charophytes (Table 1). The predominant E. fluviatile oligosaccharide co-migrated with G4G4G3G; other appreciable products included G3G (labelled 2′ in Fig. 1c) and G4G3G. An additional minor disaccharide (labelled 2 in Fig. 1c), which ran slightly more slowly than cellobiose and was found only in Equisetum digests, remains unidentified. Similar oligosaccharide patterns were obtained from all Equisetum species tested: Earvense, Efluviatile, Escirpoides, Esylvaticum, and E. × trachyodon (results not shown).

Identification of MLG oligosaccharides by alkaline peeling

The identity of the lichenase digestion products was further supported by their reaction with NaOH. This treatment causes ‘peeling’ of oligosaccharides – that is, sugar groups are removed one by one from the reducing terminus by an elimination reaction (Whistler & BeMiller, 1958; Stone & Clarke, 1992; Fry, 2000). It is known that (1→3) bonds are particularly susceptible to such peeling, the original reducing-terminal glucose moiety being released as an epimeric mixture of d-glucometasaccharinates. By studying model trisaccharides we confirmed the greater alkali lability of (1→3) than (1→4) bonds, a characteristic proposed as the basis of a method for sequencing oligosaccharides (Luchsinger & Stone, 1976) (Fig. 2). After 1–4 h in NaOH, laminaritriose (G3G3G) had produced a high yield of G3G and a trace of glucose, whereas with cellotriose (G4G4G) the rate of G4G production was ~16 times slower.

Figure 2.

Time-course for the alkaline peeling of oligo-(1→3)- and (1→4)-β-d-glucans. Laminaritriose and cellotriose were treated with 50 mM NaOH at 20°C for 10 s to 64 h. At intervals, samples were neutralized with acetic acid and analysed by thin-layer chromatography (TLC) as in Fig. 1 except that this plate was developed only once so mobilities are lower than in Fig. 1. Control, trisaccharides not treated with alkali; Lam-oligos, marker mixture of glucose, laminaritriose and laminaritetraose; Cello-oligos, marker mixture of glucose, cellobiose, cellotriose and cellopentaose.

As lichenase-generated oligosaccharides are expected to have a (1→3) bond next to the reducing terminus, they should readily lose one glucose moiety by peeling, leaving a relatively stable cello-oligosaccharide. Peeling of G4G4G4G3G, G4G4G3G, G4G3G and G3G is expected to yield cellotetraose, cellotriose, cellobiose and glucose, respectively, and these products were indeed formed by NaOH treatment of the lichenase products of Festuca and Equisetum (Fig. 1b,c; arrows). Larger oligosaccharides also underwent peeling but their products did not generally co-migrate with simple cello-oligosaccharides, supporting the idea that some of them had arisen by incomplete enzymic digestion and thus contained both a terminal (1→3) bond (alkali-labile) and a nonterminal (1→3) bond (alkali-resistant). The alkaline degradation of Equisetum disaccharide 2′ and simultaneous formation of a trace of glucose (Fig. 1c) are consistent with its being laminaribiose.

HPLC analysis of Equisetum MLG oligosaccharides

HPLC gave good resolution of MLG oligosaccharides (Fig. 3). Commercial barley-flour MLG and hemicellulose B from Avena coleoptiles, when lichenase-digested to completion (72 h), gave a series of MLG oligosaccharides. In these poalean samples, HPLC confirmed that the trisaccharide exceeded the tetrasaccharide. Smaller amounts of oligosaccharides with degree of polymerisation (DP) up to at least 13 were detectable; favoured sizes within this range were the nonasaccharide and the dodecasaccharide. Laminaribiose was an extremely minor product.

Figure 3.

High-pressure liquid chromatography (HPLC) characterization of Equisetum mixed-linkage (1→3,1→4)-β-d-glucan (MLG). Markers were (a) glucose and cello-oligosaccharides (C2–C10), and (b) glucose and laminari-oligosaccharides (L2–L4). Lichenase digestion products of (c) barley flour MLG, (d) oat coleoptile hemicellulose B, (e) Equisetum arvense hemicellulose B, and (f) Equisetum sylvaticum hemicellulose B. For (c–f), the polysaccharides were heated at 100°C and then digested with lichenase for 72 h (black profiles); grey profiles show enzyme-free controls. G3–G12, estimated degree of polymerization of lichenase-generated oligosaccharides.

The tri- and tetrasaccharides from Equisetum had the same retention time as those from cereals, supporting their identity (Fig. 3). Equisetum arvense and E. sylvaticum were selected for Fig. 3 because they were richest and poorest in MLG, respectively, among the five Equisetum spp. tested; the others (E. fluviatile, E. scirpoides and E. ×trachyodon) gave HPLC profiles intermediate between those of E. arvense and E. sylvaticum (data not shown). In each case, the tetrasaccharide (G4G4G3G) was very clearly the major product. Thus the HPLC data confirmed that Equisetum MLGs yield a much higher proportion of G4G4G3G than do poalean MLGs, and that minor products include laminaribiose and G4G3G. Larger oligosaccharides up to DP9 were also present. As in the Poaceae, the nonasaccharide itself was a favoured size; however, unlike in the Poaceae, no dodecasaccharide was detectable (Fig. 3). A nonasaccharide stable for 72 h in the presence of lichenase could arise from the sequence ---G3G4G4G4G4G4G4G4G4G3G4G---, the underlined bonds being cleaved by lichenase.

The fact that laminaribiose was much more abundant in Equisetum digests than in cereal digests suggests that this disaccharide was not formed as a result of a contaminating enzyme in the lichenase preparation but that it indicated the presence of a sequence (---G3G4G3G4G---) that is much more abundant in Equisetum than in cereals.

MLG as a structural component of Equisetum cell walls

To test whether Equisetum MLG may be strongly hydrogen-bonded to cellulose, as proposed for the Poaceae (Wada & Ray, 1978; Carpita et al., 2001), we attempted to extract it using agents other than NaOH, the conventional hemicellulose extractant. DMSO extracted c. 15, 5 and 12% of the MLG from E. fluviatile, Z. mays and F. arundinacea AIR, respectively (Table 2). Boiling water extracted little additional MLG from E. fluviatile or Z. mays and only a further 6% from F. arundinacea. In all three species a substantial majority of the MLG was left firmly wall-bound after sequential extraction with DMSO and boiling water, suggesting that it is tightly associated with the cellulose and may act as an inter-microfibrillar tether.

Table 2.  Differential extraction and lichenase-digestion analysis of mixed-linkage (1→3,1→4)-β-d-glucans (MLGs) from a horsetail, a cereal and a grass
PlantFractionLichenase digestion products (µg mg−1 AIR)Molar ratio, di-:tri- :tetra-saccharide
Laminaribiose, G3GMLG trisaccharide, G4G3GMLG tetrasaccharide, G4G4G3GSum of 3 oligosaccharides
  1. Alcohol-insoluble residue (AIR) was prepared from May-harvested stems of Equisetum fluviatile and from suspension-cultured cells of the cereal Zea mays and the grass Festuca arundinacea. Each AIR sample was treated with DMSO at 20°C, and then with water at 100°C, and finally the insoluble pellet was collected. Both extracts were dialysed, and the two dialysates as well as the insoluble pellets were lichenase-digested and analysed by high-pressure liquid chromatography (HPLC; as in Fig. 3). The MLG oligosaccharides were quantified on the assumption that they gave the same detector response factor as laminari-oligosaccharides of the same degree of polymerization. DMSO, dimethyl sulphoxide.

Equisetum fluviatileDMSO extract0.770.603.494.9188 : 100 : 439
Hot water extract0.0540.0330.280.36244 : 100 : 636
Insoluble pellet2.542.2822.026.8164 : 100 : 730
Total3.362.9125.832.0170 : 100 : 670
Zea maysDMSO extract0. : 100 : 61
Hot water extract0.000.0570.0440.100 : 100 : 57
Insoluble pellet0.0124.623.908.50.4 : 100 : 64
Total0.0124.954.169.10.4 : 100 : 64
Festuca arundinaceaDMSO extract0.0792.641.884.64.4 : 100 : 54
Hot water extract0.001.540.922.50 : 100 : 45
Insoluble pellet0.09218.812.231.10.7 : 100 : 49
Total0.1723. : 100 : 49

The tri:tetrasaccharide ratio of F. arundinacea MLG was ~2 : 1 in all three fractions (the two extracts and the final pellet). The ratio was also consistent (~1.6 : 1) in the three Z. mays fractions. However, differential extraction did effect some qualitative fractionation in the case of E. fluviatile MLG – the extractable fractions were appreciably richer in trisaccharide repeat units than was the final pellet. This indicates that E. fluviatile MLGs comprise a spectrum of polysaccharides varying subtly in linkage ratios.

Quantification of MLG in Equisetum fluviatile

The total content of MLG, estimated as the sum of lichenase-generated di + tri + tetrasaccharide, amounted to c. 3.2, 0.9 and 3.8% (w/w) of the AIR of E. fluviatile stems, Z. mays cultures and F. arundinacea cultures, respectively (Table 2). Because this analysis disregards the domains of MLG that yield oligosaccharides larger than the tetrasaccharide, the total MLG must somewhat exceed these quoted percentages.

Presence of xyloglucan in Equisetum

Samples of hemicellulose B were incubated with Driselase, which digests xyloglucan to isoprimeverose (α-d-xylopyranosyl-(1→6)-d-glucose) plus monosaccharides (Fry, 2000; Thompson & Fry, 2001; Popper & Fry, 2003). Equisetum hemicellulose gave a moderate yield of isoprimeverose, comparable to that of many other land plants (Fig. 4b, and data not shown), but only small amounts of xylobiose, indicating a low xylan content. It yielded fucose, galactose and glucose (all three of which could arise partly from xyloglucan) and large amounts of mannose. It also produced an unidentified disaccharide (eluting ~1.5 min before isoprimeverose) that was practically undetectable in angiosperms but did occur in hemicellulose digests from gymnosperms, ferns, Psilotum, lycopodiophytes and bryophytes (data not shown). The high isoprimeverose:xylobiose ratio observed in Rosa is typical of dicot primary cell walls. The high isoprimeverose:xylobiose ratio observed in Z. mays is typical of cereal cell cultures (Kerr & Fry, 2003), rather than mature cereal leaves.

Figure 4.

High-pressure liquid chromatography (HPLC) of xyloglucan-derived oligosaccharides from a horsetail, a cereal and a dicotyledon. The hemicellulose B fraction from stems of two Equisetum species and from cell-suspension cultures of maize (Zea mays) and rose (Rosa sp.) were digested with (a) xyloglucan endoglucanase or (b) Driselase, and the products analysed by HPLC. The sample injected was equivalent to (a) 400 µg or (b) 90 µg of hemicellulose B. Different elution programmes were used for (a) and (b). In (a) the solid- and open-headed arrows indicate the positions of XXXG and XXFG, respectively; in (b) solid-headed arrows indicate isoprimeverose. The markers for (a) were obtained by xyloglucan endoglucanase (XEG) digestion of tamarind xyloglucan. Ara, arabinose; C2, cellobiose; Fuc, fucose; Gal, galactose; GalA, galacturonic acid; Glc, glucose; GlcA, glucuronic acid; IP, isoprimeverose; Man, mannose; Rha, rhamnose; Rib, ribose; Xyl, xylose; X2, xylobiose.

Additional samples of hemicellulose were incubated with XEG, which digests xyloglucans to characteristic oligosaccharides (Pauly et al., 1999). In the dicot Rosa, the main xyloglucan oligosaccharides were XXFG and XXXG, as has been well established (McDougall & Fry, 1991). (For explanation of the abbreviated nomenclature of xyloglucan oligosaccharides, see Fry et al. (1993).) In the cereal Z. mays, the predominant xyloglucan oligosaccharide had a retention time of 8.9 min and may be XXGG (Gibeaut et al., 2005). Equisetum hemicellulose gave moderate yields of xyloglucan oligosaccharides, comparable to those of many other land plants including Z. mays and Rosa (Fig. 4a). The two Equisetum species gave oligosaccharide profiles very similar to each other but markedly different from those of all other taxa tested, suggesting that Equisetum xyloglucan has unique structural features. The Equisetum digests gave a peak co-eluting with authentic XXXG, and a pronounced shoulder with the same retention time as XXFG. However, the putative XXXG peak of Equisetum also had an asymmetry suggesting the presence of an additional oligosaccharide eluting slightly earlier. Several other nonstandard oligosaccharides were also abundant, especially one that eluted at 11.4 min, which was scarce or absent in digests from angiosperms, gymnosperms, leptosporangiate ferns, Psilotum, lycopodiophytes and bryophytes (Fig. 4a and data not shown).


MLG is present in Equisetum

The discovery in Equisetum of an endotransglucosylase activity that favours MLG as its donor substrate (Fry et al., in press) raised the intriguing possibility that MLG or a similar polysaccharide may be present in this genus. Among vascular plants, MLG has so far been reported only in the Poales. Popper & Fry (2004) did not detect MLG in Equisetum debile. Although E. debile was not available for the present work, we did detect MLG in five other Equisetum species.

Lichenase digestion of poalean MLG yields a trisaccharide (G4G3G; majority) and a tetrasaccharide (G4G4G3G) (Stone & Clarke, 1992; Roubroeks et al., 2000). The tri:tetrasaccharide ratio is ~2 : 1 in oat (Avena sativa) flour, ~3 : 1 in barley flour, and up to 4.5 : 1 in wheat (Triticum aestivum) flour (Li et al., 2006). Contiguous (1→3) bonds do not occur, and evidence for small quantities of the sequence G3G4G3G4G (which on lichenase digestion would yield G3G) has been the subject of debate (Roubroeks et al., 2000). The trisaccharide and tetrasaccharide together constitute ~90–92% of wheat flour MLG; of the remaining 8–10%, a nonasaccharide often predominates (Li et al., 2006), as observed by us in barley flour and cultured Festuca cells (Fig. 3). We also detected traces of G3G in digests of Festuca AIR. The MLG of lichens (‘lichenan’) yields essentially only the trisaccharide G4G3G on lichenase digestion (Stone & Clarke, 1992).

In lichenase digests of all Equisetum species tested, the tetrasaccharide G4G4G3G was by far the major product, a clear difference from poalean MLGs. The trimer G4G3G was a relatively minor product, often exceeded by the dimer, G3G. Another product from Equisetum was a nonasaccharide – also obtained from poalean MLGs and proposed to arise from a ‘cellulose-like’ domain with eight contiguous (1→4) bonds. However, a dodecasaccharide in the Poaceae digests, presumably indicating eleven contiguous (1→4) bonds, was not obtained from Equisetum. Thus, Equisetum possesses MLG, but this polysaccharide differs significantly from the MLGs of the Poaceae and the lichens (Table 3).

Table 3.  Summary of predominant sequences detected in mixed-linkage (1→3,1→4)-β-d-glucan (MLG) of Equisetum spp. and the Poaceae
SequenceaRelative abundance of sequence in MLG of
  • The table gives a consensus derived from analysis of several species in each taxon; ++++, major component; +, minor component; ±, detectable in some samples only.

  • a

    Underlined, bond cleaved by lichenase. DP, degree of polymerization of the ‘cello-oligosaccharide’ domain flanked by two (1→3) bonds.

---G3G4G3G4G--- (DP 2)++±
---G3G4G4G3G4G--- (DP 3)++++++
---G3G4G4G4G3G4G--- (DP 4)++++++
---G3G4G4G4G4G3G4G--- (DP 5)++
---G3G4G4G4G4G4G3G4G--- (DP 6)±+
---G3G4G4G4G4G4G4G4G4G3G4G--- (DP 9)++
---G3G4G4G4G4G4G4G4G4G4G4G4G3G4G--- (DP 12)+

Quantitatively, MLG constitutes a relatively high proportion of the Equisetum cell wall (32 mg g−1 dry weight in E. fluviatile). Primary cell walls are estimated to have a total polysaccharide content of ~40% on a wet-weight basis (Monro et al., 1976); thus, MLG at 32 mg g−1 wall dry weight corresponds to ~1.2% of the wet weight, or roughly 10 mg ml−1.

MLG is considered to adopt a worm-like extended coil conformation (Buliga et al., 1986; Roubroeks et al., 2000). If it is not hydrogen-bonded to cellulose, it can be dissolved in hot water to form dilute solutions; but cooling, storage or concentration can cause the solutions to gel. Gelling may be a result of inter-chain hydrogen-bonding between the ‘cellulose-like’ domains (Woodward et al., 1983) and/or contiguous trisaccharide units (Böhm & Kulicke, 1999). In agreement with the latter suggestion, poor solubility and rapid gelation are characteristic of poalean MLGs that possess a high tri:tetrasaccharide ratio, such as those of wheat flour (Li et al., 2006), in which the rarity of tetrasaccharide units ensures that trisaccharide units will often be contiguous. MLG has been reported to bind noncovalently with plant cell wall arabinoxylans (Wada & Ray, 1978; Izydorczyk & MacGregor, 2000). MLG, like other hemicelluloses such as xyloglucan, can hydrogen-bond to cellulose. Indeed, G4G4G3G and larger oligosaccharides of MLG bind so firmly to filter paper (cellulose) that they are not amenable to paper chromatography: they remain firmly bound at the origin or produce an indistinct streak rather than a discrete spot (data not shown).

The tendencies of MLG to self-aggregate and bind to other hemicelluloses and cellulose point to a potential major architectural role for MLG in the Equisetum cell wall – for example, as an inter-microfibrillar tether, as proposed earlier for xyloglucan in dicots and for MLG in grasses (Labavitch & Ray, 1978; Wada & Ray, 1978; Fry, 1989; Hayashi, 1989). Sørensen et al. (in press), who have independently detected MLG in Equisetum arvense, found it to be more abundant in older regions of the stem, suggesting a possible role in wall strengthening in mature tissue.

Evolutionary significance of MLG occurrence in Equisetum

MLG had been thought to be restricted, among vascular plants, to the Poales. Its discovery in Equisetum was therefore unexpected. Because Equisetum diverged from its closest living relatives (other early-diverging ‘ferns’) > 380 Ma (Bell & Hemsley, 2000), there is no possibility that the discovery of MLG in Equisetum indicates a close relationship to the Poaceae (an angiosperm family that diverged c. 65 Ma; Kellogg, 2001). Rather, it seems that the ability to synthesize MLG has evolved at least twice, independently, in the history of vascular plants. That this is a rather facile evolutionary step is suggested by the successful genetic transformation of the dicot Arabidopsis to enable MLG synthesis by insertion of a single gene encoding a cellulose synthase-like (CslF) enzyme (Burton et al., 2006). It is also suggested by the finding of an arabinosylated MLG-like polysaccharide in a liverwort (Popper & Fry, 2003). The possible occurrence of MLGs in additional (maybe quite narrow) plant and algal groups will be an interesting topic for future research.

Although there is no close evolutionary relationship between the Poales and Equisetum, there are several functional similarities between grasses and horsetails, including an often ‘wiry’ aerial growth habit, basal growth of the internodes, most of the roots arising adventitiously, the presence of MLG, and a high silica content. These shared features must have been arrived at by convergent evolution.

Correlation between MLG and silica deposition

Equisetum epidermal cell walls contain exceptionally high concentrations of silica (Currie & Perry, 2007; Sapei et al., 2007). ‘Knobs’ on the ridges of E. hyemale stems contain up to 33% silicon in the form of amorphous, colloidal silica containing numerous silanol groups (Si–OH; as found in silica-gel) and little organic matter, whereas the remaining epidermal cell walls have uniform silicification in the form of colloidal sheets in which silanol groups appear to be largely replaced by hydrogen bonds to polysaccharides (Sapei et al., 2007).

The other land plants famous for accumulating silica are in the Poales (Raven, 1983; Hodson et al., 2005; Currie & Perry, 2007). Both Equisetum and grasses are MLG-rich, raising the intriguing possibility that MLG serves a key role in the poorly understood mechanisms of cell wall silicification. The abundance of MLG in the epidermal walls of Equisetum (Sørensen et al., in press) supports this idea, as the epidermis is heavily silicified. A role for polysaccharides in silica polymerization has long been suspected, and cello-oligosaccharides can regulate the growth of silica particles (Harrison & Lu, 1994). Furthermore, a shift from sheet-like to globular silica deposition during the development of Phalaris (canary grass) trichomes was closely correlated with a large increase in MLG biosynthesis (Perry et al., 1987). These observations strongly suggest a role for MLG in regulating the location and/or quality of silica deposited.

Some algae also have a high silica content, especially in the Bacillariophyceae (diatoms) and Synurophyceae (formerly included in the Chrysophyceae). These organisms have not been shown to possess MLG, although some of them do have (1→3,1→6)-β-d-glucans, mainly in the vacuoles (Storseth et al., 2006). Some other algae, not noted for a high silica content, do possess MLG – for example Monodus subterraneus (in the Xanthophyceae; Ford & Percival, 1969) and possibly Peridinium westii (a dinoflagellate; Nevo & Sharon, 1969). Further studies to test whether silicified algae have any wall polysaccharides resembling MLG would be valuable.

Xyloglucan–MLG co-occurrence in Equisetum

The discovery in Equisetum of an endotransglucosylase that favours MLG as its donor substrate and xyloglucan as acceptor substrate (Fry et al., in press) suggested that MLG and xyloglucan would co-occur in Equisetum cell walls. However, immunocytochemical observations (Sørensen et al., in press) suggested that these two polysaccharides were concentrated in different tissues. Our analyses of Equisetum shoot total hemicellulose B revealed moderate levels of isoprimeverose and xyloglucan oligosaccharides, comparable to those in many other land plants (Popper & Fry, 2004). The xyloglucan oligosaccharide profiles (Fig. 4, and data not shown) suggest that Equisetum xyloglucans are rather distinct from those of all other land plants, so the possibility exists that immuno-techniques may not show them up efficiently. The total xyloglucan content of Equisetum stems may be similar to that in most other vascular plants. On balance, we suggest that MLG and xyloglucan co-occur in the cell walls of Equisetum stems.


We thank the BBSRC (UK) for funding this research, and Sentil Kumar Gandi and Yoshitaka Hiruma for technical assistance. We also thank Zoë Popper, Lorna MacKinnon and Peter Wylie for collecting some of the plants.