Author for correspondence: A. Ashford Tel: +61 293852068 Fax: +61 29385 1558 Email: : email@example.com
• Hair roots of Woollsia pungens are shown to have thick-walled epidermal cells, a feature found in a small number of other species within the Epacridaceae. Hair roots otherwise had a structure typical of the Ericales.
• Ultrastructural, immunocytochemical and histochemical techniques were used to investigate the structure and composition of these thick-walled epidermal cells.
• The thick walls were multilamellate with a helicoidal arrangement of microfibrils typical of a secondary cellulosic wall. Staining techniques revealed a relatively high abundance of β-glucans; these were not β 1–3 linked and there was no detectable protein. Galactose side-chains were abundant but not mannose or glucose side-chains. The wall contained a pH-dependent net negative charge. Although apparently rich in COOH groups the thick wall did not react, or only minimally, with the monoclonal antibodies JIM5 and JIM7, testing for nonesterified and methyl-esterified pectins, respectively; this contrasted with the strong positive reaction in the cortical and stelar cells. In epidermal cells colonized by mycorrhizal fungi the thick wall had additional layers of spongy appearance with many interconnected, irregular patches containing dispersed material. Colonized cells retained their integrity longer than noncolonized cells.
• The thick wall might be important in long-term survival of infected cells and the low levels of pectin might control mycorrhizal endophyte infection.
Members of the two very closely related families Ericaceae and Epacridaceae (Kron, 1996) show floristic and morphological similarities and a marked preference for nutrient-poor, acidic heathlands or sandplains (Read, 1996). So far, virtually all heathland plants of the Ericaceae and all but two members of the Epacridaceae have been found to produce extremely fine hair roots (30–100 µm in diam.) which develop endomycorrhizal associations of the ericoid type. These are implicated in the acquisition of organic, often intractable, forms of nitrogen (Bell et al., 1994; Read, 1996; Smith & Read, 1997). Hair roots are generally considered to be ephemeral and show distinctive patterns of seasonal development according to habitat and hydrological conditions (Read, 1996; Smith & Read, 1997). The finest, ultimate, hair roots have a very small stele surrounded by only two layers of cortical cells and an epidermis. The epidermal cells become colonised by fungal hyphae to form coils and an ascomycete is usually involved (Bonfante-Fasolo & Gianinazzi-Pearson, 1979; Read, 1983; Allen et al., 1989; Ashford et al., 1996). Apart from their ability to form mycorrhizal associations, there is little understanding of the physiology of these specialised hair roots in nutrient and water uptake, though the suberisation of the cortex and the small diameter of the xylem elements indicate low to negligible rates of water transport (Allaway & Ashford, 1996).
In a small number of species within the Epacridaceae, the epidermal cells develop a thick wall. These include Dracophyllum secundum R. Br. (Allen et al., 1989, J. G. Duckett, pers. comm.), Acrotriche serrulata (Labill.) R. Br., Astroloma spp., Lysinema ciliatum R. Br., Lysinema elegans Sonder, Needhamiellapumilo (R. Br.) L. Watson, Oligarrhena micrantha R. Br. and Woollsia pungens (Cav.) F. Muell. (Reed, 1987; Ashford et al. 1996). In L. ciliatum from Western Australia this wall was shown to have an unusual structure, there being several distinct wall layers some of which appeared spongy (Ashford et al., 1996). Thick-walled cells were preferentially colonised by the mycorrhizal fungi and were longer lived. They were also easily shed from the surface of the root and it was suggested that in nature they may be released into the rhizosphere and that the fungus therein may survive until favourable conditions return. They, therefore, could act as centres of infection from which the fungus is able to colonise newly emerging roots. There are several potential roles for the thickened wall of the epidermal cell. It may serve a protective function, influence the water relations within the cell and/or provide a substrate for the fungus to enable hyphal outgrowth (Ashford et al., 1996).
Here we have examined hair roots of Woollsia pungens, a member of the Epacridaceae from Eastern Australia, and shown that they similarly have thick-walled epidermal cells which are colonised by ascomycete mycorrhizal fungi and can also become detached from the root. We have investigated the structure and composition of this thick-wall in W. pungens, using a range of ultrastructural, immunocytochemical and histochemical techniques to explore its potential role in hair roots and the mycorrhizal symbiosis.
Materials and methods
Plants of Woollsiapungens (Cav.) F. Muell. were collected from Jennifer Street Reserve, La Perouse, NSW, Australia (33°59′S, 151°14′E) in late November 1997, and Cherrybrook, NSW, Australia (33°46′S, 151°4′E) in early August 1998. Soil was carefully washed away from the root system and the hyaline, or lightly coloured hair roots, were removed. Some were examined whole and unfixed, whilst others were processed either for transmission electron microscopy or immunocytochemistry. Data were compiled from more than 30 roots from 4 plants. There were no obvious differences in structure or histochemistry of hair roots collected at the 2 different sites.
For transmission electron microscopy, roots were fixed in 1% paraformaldehyde plus 2% glutaraldehyde in either 0.1 M PIPES buffer (pH 7.2), or 0.1 M cacodylate buffer (pH 6.98) for 1–2 h at room temperature (RT) followed by 15 h at 4°C. They were then rinsed with 3 changes of the buffer, postfixed in 2% (aq) osmium tetroxide for 1 h, rinsed in double distilled water and dehydrated through a graded ethanol series (all at RT), a technique modified from Hawes (1994). They were gradually infiltrated with medium grade LRWhite resin (London Resin Co. London, UK) over 6 d at 4°C and polymerized for 18 h at 57°C. Ultrathin sections were cut, collected on formvar-coated copper slot grids, stained with 2% (aq) uranyl acetate (16 min) and Reynolds (1963) lead citrate (10 min), and examined with an Hitachi 7000 transmission electron microscope at 75 kV.
For lectin and JIM antibody staining, roots were fixed in 1% paraformaldehyde plus 1% glutaraldehyde in 0.05 M PIPES buffer (pH 7.2) with a drop of Tween 20 for 1.5–2 h at RT, rinsed 3 times with PIPES buffer, dehydrated with a graded ethanol series at 4°C and infiltrated in either several changes of degassed butyl-methyl methacrylate or LRWhite resin for 8 d at 4°C. Roots infiltrated in LRWhite resin were polymerized as above. Roots infiltrated in butyl-methyl methacrylate were polymerized in fresh resin overnight at RT under UV light in a nitrogen atmosphere (Gubler, 1989; Baskin et al., 1992). Sections (0.5–2.0 µm thick) were collected on glass slides either coated with 3% 3-aminopropyltriethoxysilane (APES, Sigma A3648) dissolved in 100% acetone (Stoddart & Jones, 1997), or 10% poly L-lysine (Sigma P8920). Sections were individually ringed by a PAP PenT (Daido Sangyo Co. Ltd. Japan) for application of the lectins or antibodies. Butyl-methyl methacrylate was removed from sections by 100% acetone prior to lectin and JIM antibody staining.
Sections (0.5 µm thick) were treated with a range of stains according to O’Brien and McCully (1981) unless otherwise stated. These were: amido black 10 B 1% in 7% acetic acid (Fisher, 1968); toluidine blue (C.I. 52040 BDH) at 0.05% in 0.012 M NaHCO3 (pH 9.2), or 0.5% in 0.1 M sodium acetate buffer (pH 4.4) or 0.05% in 0.025 M HCl-KCl buffer (pH 1.0); alcian blue 8GX (C.I. 74240 Sigma) at 1% w/v in 0.1N HCl (pH 1.0), or 1% w/v in 3% acetic acid (pH 2.5), or 0.1% w/v in 0.05 M sodium acetate buffer (pH 5.7 with or without the addition of MgCl2 at the following concentrations −0.1 M, 0.2 M, 0.5 M, 0.8 M and 1.0 M) (Pearse, 1968) with staining times of 4, 15 and 30 min; Periodic acid-Schiff (PAS) reagents, blocked for 30 min with 2,4-dinitrophenyl hydrazine in 15% acetic acid as the aldehyde blocking agent for 30 mins. Calcofluor White M25 (C.I. 40622 Sigma), 0.023% w/v in distilled water; aniline blue (C.I. 42755 BDH) at 0.05% in 0.067 M K-phosphate buffer (pH 8.5) with a control of pH 8.5 buffer alone to determine autofluorescence.
The following fluorescein isothiocyanate (FITC)-labelled lectins were tested on hair root sections using a method modified from Stoddart & Jones (1997): peanut agglutinin (PNA) from Arachishypogaea (Sigma L-7381), Bandeiraea simplicifolia agglutinin (BS-I) (Sigma L-2895) and Concanavalin A (ConA) from Canavalia ensiformis (Sigma C-7642). A series of titre runs was first carried out to determine the most appropriate lectin concentration, buffer pH, cation composition and incubation temperature. The schedule below gave the best and most consistent staining. Sections were first incubated with 50 µl of 20 mM Tris buffered saline (TBS) containing 2% bovine serum albumin (Sigma A-7030), 12 mM CaCl2, 2 mM MgCl2 and 2 mM MnCl2 (blocking solution) for 30 min. Then a solution of 150 µg/ml FITC-lectin was added to make a final concentration of 75 µg/ml lectin. Sections were incubated for 60 min, rinsed in TBS + cations and examined in TBS + cations with Citifluor (City University, London cited Oparka & Read, 1994) added to a ratio of 1 : 1. Reactions for PNA and B. simplicifolia-BSI were carried out at pH 7.8, and for ConA at pH 8.0. All incubation steps were carried out at 37°C in a moist environment.
Controls consisted of either replacing the lectin with FITC-BSA (Sigma A-9771), or blocking the lectin with the appropriate sugar before applying it to the sections. ConA was blocked with 125 mM D-glucose plus 630 mM methyl α-D-mannopyranoside (Sigma M-6882), while PNA and B. simplicifolia-BSI lectins were blocked with 200 mM D(+) galactose (Sigma G-0750).
Unfixed, whole hair roots were also used for testing with the FITC-lectins.
JIM5 and JIM7, two monoclonal antibodies which react with epitopes of un-esterified and methyl-esterified pectins, respectively (Knox et al., 1990) were also tested. Sections were first blocked with 25 µl of 10 mM phosphate buffered saline containing 2% BSA (PBS/BSA), pH 7.46 for 30 min. The blocking agent was carefully removed with filter paper and each section covered with 50 µl of a solution of JIM7 or JIM5 diluted with PBS/BSA to a concentration of 1 : 7 or 1 : 1 for JIM7 and 1 : 10 for JIM5. The slides were incubated for 60 min, rinsed 3 times with PBS and excess PBS was removed with filter paper. Each section was then covered by 50 µl of antirat IgG-FITC (Sigma F-6258) at a dilution of 1 : 40. The sections were incubated for 60 min, rinsed with PBS then examined in a 1 : 1 mixture of buffer and Citifluor. Control sections were treated in the same way but the primary antibody step was omitted. All reactions were carried out in a moist environment at 25°C.
Field emission scanning electron microscopy (FESEM)
To determine the distribution of sulphur throughout the epidermal walls, semi-thin (0.5 µm) transverse-sections of nonosmicated hair roots (fixed as for immunocytochemistry and embedded in LRWhite resin) were collected on formvar-coated copper slot grids and examined by an energy dispersive X-ray system attached to an Hitachi S-4500-II Field Emission Scanning Electron Microscope. Spectra for the cell walls, cytoplasm and resin were collected.
Brightfield images were photographed on a Leitz Orthoplan microscope (Ernst Leitz, Wetzlar, Germany) using Kodak 64T Ektachrome film. Calcofluor white and Aniline Blue stained sections were examined in the epifluorescence mode using the filter combination BP 340–380 nm, RKP 430 nm, LP 430 nm. The lectin and JIM antibody stained sections were examined by epifluorescence microscopy (Earlzeiss, Microscope Diusion 73446, Oberkochen, Germany) using a Zeiss Axiophot microscope using the filter combination BP450–400 nm, FT 510 nm, LP 515–585 nm. Images were captured on Kodak Elite II 400 ISO.
Woollsia pungens hair roots are 30–100 µm in diameter and consist of a single-layered epidermis, a two-layered cortex differentiated into an outer exodermis with a suberized lamella and an inner endodermis with well developed Casparian strips, and a small stele. In the finest hair roots (approx. 30 µm in diameter), there are 4–5 cells in each of the epidermal and cortical cell layers, a single xylem tracheid, a single sieve cell, one companion cell and two or three pericycle cells (Fig. 1a). In larger roots (approx. 75 µm in diameter) there were 7 cell profiles in each of the epidermal and cortical layers, and two obvious xylem ridges alternating with two phloem strands, indicating the roots are diarch (Fig. 1b).
The root apex comprises a small apical meristem overlain by a very small root cap consisting of a few cells covered by a thick layer of mucilage (slime) which was metachromatic (pink) with toludine blue (Fig. 1c) and ConA positive (not shown) indicating mannose and glucose side chains. Epidermal and cortical cells are differentiated close to the apex (Fig. 1c) and the developing vacuoles in the epidermal, exodermal and endodermal cells contain electron-opaque deposits, likely to be phenolic compounds on the basis of their staining with toluidine blue (Fig. 1b–f).
All epidermal cells examined had thickened walls. The inner tangential wall is the thinnest region, gradually increasing from approx. 0.2 µm in the centre to approx. 0.5 µm at the radial wall junction. The radial walls gradually increase in thickness from approx. 0.5 µm at the junction with the inner tangential wall to approx. 2.0 µm at the junction with the outer tangential wall. The outer tangential walls range in thickness from 2.0 to 3.5 µm. Some epidermal cells appear to be empty and showed various degrees of collapse (Figs 1a,d,e, 2a), whilst others contain fungal profiles (Figs 1a,d–f, 2a,b). Colonised cells retain their contents and do not collapse. The epidermal cells tend to separate along the middle lamella in older living roots and are easily detached (not shown). Separation of epidermal cells along the middle lamella of the radial walls is also seen in embedded material, possibly as a result of processing (Fig. 1b).
The host cell cytoplasm of colonised cells, though not well fixed, contains the usual organelles, suggesting that it was not moribund at the time of fixation. The two colonised cells in Fig. 2(a) both have a large central vacuole containing deposits of electron-opaque material surrounded by a thin layer of peripheral cytoplasm. A layer of electron-opaque material indicates the position of the tonoplast. Profiles of fungal hyphae which appeared to be within the vacuole are each surrounded by a thin layer of cytoplasm suggesting that the vacuole has been invaginated by the infection (Fig. 2a,b). By contrast, collapsed cell profiles contain varying amounts of electron-opaque material, but no obvious organelles or other recognizable cellular structures (Fig. 2a,c).
There are also obvious differences between the walls of collapsed, un-colonized epidermal cells and colonised cells. The thickened wall of un-colonised cells are multilamellate and showed a helicoidal structure throughout (cf. Roland et al., 1987; Mizuata et al., 1994), with alternating bands of microfibrils showing arc-like patterns. This is best illustrated in Fig. 3(a) where there is least displacement in the innermost wall region. Here the lamellae are more or less flat and show the typical arc-shaped pattern of the microfibrils arranged helicoidally in alternating layers (Fig. 3a). This pattern becomes somewhat displaced with increasing distance from the inner wall surface, so that the lamellae (and their microfibrils) undulate in a pattern which generally follows the undulations of the outer part of the wall. This effect is presumably due to release of tensions in the wall as the cell collapses. Overlying the wall region of obvious microfibrillar texture is a thin amorphous wall layer with a more electron-opaque band at its outer edge. In some cells this electron-opaque band appears as a distinct layer at the boundary separating the outer surface of the wall from the overlying mucilaginous layer (Fig. 2a,c), but in others (presumably depending on staining and plane of section) this layer is more diffuse and less distinct (compare Fig. 3a,b with 2a,c). The radial walls also exhibit lamellae with electron-opaque areas but the arc-shaped patterning is blurred (not shown) as is also the case in any part of the wall that is deeply folded (Fig. 2c).
In colonised epidermal cells, walls are more complex. The helicoidal wall microfibril pattern is again found, usually in the outermost region of the wall which is multilayered (Figs 2a, 3c). Additional wall layers are also present (Figs 2a, 3d). These have a spongy appearance with interconnected variously shaped, electron-lucent areas (Figs 2a, 3d, 3e), containing varying amounts of material (compare Fig. 3d with 3e). Wall microfibrils are not visible. In many infected cells there appear to be several partially superimposed spongy layers within a more conventional multilamellate wall (Figs 2a, 3d).
Mucilage is retained over most of the root surface in most sections taken from differentiated root regions (Figs 2a,c, 3a,b). It varies in thickness and texture and is often overlain by a very thin, electron-opaque layer (Figs 2c, 3a,b). The mucilage layer around the many hyphal profiles on the root surface (Fig. 2a) is thicker than elsewhere and it occasionally has debris adhering to the surface (Fig. 2a). Surface hyphae all have cell contents and were obviously alive at the time of fixation. The fungi involved belong to the Ascomycotina as indicated by their simple septa and Woronin bodies (Fig. 4).
Histochemistry of the epidermal cell walls
There was essentially no difference between osmicated and nonosmicated material except for a slight increase in the intensity of the staining reaction with some dyes in the latter. Table 1 summarizes the staining reactions of the thick cell wall in nonosmicated material. There was no reaction with amido black 10B (napthol blue black) a general stain for protein. The thick wall stained for acidic polysaccharides with toluidine blue and alcian blue. Toluidine blue stained the helicoidal wall region an intense, deep pink at pH 9.2 (Fig. 1d), a light to dark pink at pH 4.4 (Fig. 1e,f), but did not stain at pH 1.0 (not shown). In some cells, the helicoidal wall region had a slight bluish reaction at pH 4.4 compared with adjacent colonized cells, indicating incorporation of phenolic compounds (Fig. 1e). Spongy wall areas stained a more intense pink than helicoidal regions (Fig. 1d,f). Any overlying mucilage appeared as a very thin blue/purple layer over the root surface (Fig. 1d–f). Alcian blue at pH 5.7 and pH 2.3, stained helicoidal wall regions a light blue-green and there was a thin layer of blue-staining material over both the outer and inner surfaces of the wall. At pH 1.0 alcian blue stained the walls a pale blue with a more intense blue layer over the outer and inner surfaces (not shown). Addition of MgCl2 at 0.1 M and 0.2 M to the dye before staining resulted in a less intense blue staining and no staining occurred at 0.5 M MgCl2 and above. All wall regions stained with the PAS reaction and again the spongy wall stained more intensely than the helicoidal wall regions (Fig. 1a). Both the helicoidal and spongy wall regions also stained with Calcofluor White M2R (Fig. 5a). There was no reaction with aniline blue, although the sieve areas of the phloem and some parts of the hyphal profiles in the colonised cells were stained (Fig. 5b). None of the cell walls were labelled with FITC-Con A (not shown). FITC-peanut agglutinin (PNA) stained the helicoidal wall regions in epidermal cells strongly whilst the walls of the cortical and stelar cells did not react (Fig. 5c). Staining was patchy. There was no reaction in the controls and the only autofluorescence was from the vacuoles (Fig. 5d). The FITC-BS-I lectin also stained the thick wall of the epidermal cells (Fig. 5e–h). Comparison of Fig. 5(g) with the DIC image of the same section (Fig. 5h) confirms that the lectin binds only to the epidermal walls. Figure 5(e) shows a colonized cell where both helicoidal and perforated regions have reacted. Staining was again patchy. Again controls showed no staining and there was no autofluorescence from the cell walls (Fig. 5f).
Table 1. Staining reactions of the thick cell wall in epidermal cells of Woollsia pungens hair roots
Amido Black lOB
Very Deep Pink
pH 1.0 – O. IN HCl
pH 2.5–3% acetic acid
No added Mg C12
Pale greeny blue
0. 1 M M-Cl,
Very very pale blue
0.2 M Mg C12
Very very pale blue
0.5 M Mg CL,
0.8 M Mg C12
1.0 M Mg Cl,
– Deep Pink
Cal cofluor White –
Strong Positive Reaction
Strong positive Reaction
Aniline Blue H 8.2
Strong Positive Reaction
Strong Positive Reaction
JIM5/FITC-anti rat IgG
JIM7/FITC-anti rat IgG
No Reaction but a fine trace of lines and spots does occur in most cells examined
The pattern of staining with JIM5 antibodies was different. Comparison of the JIM5-FITC labelling (Fig. 6a) with the DIC image (Fig. 6b) of the same section, shows that the epidermal cell walls did not react but the cell walls of the exodermis and some cells of the stele were labelled. Control sections showed no staining (Fig. 6c). Sections tested with JIM7 antibodies again showed intense labelling in the cortical and stele cell walls but only a faint patchy reaction in the epidermal cell walls (Fig. 6d,e). Control sections (where the primary antibody was omitted) did not show any FITC fluorescence (not shown).
Analysis of the thick region of epidermal cell walls by energy dispersive X-ray spectroscopy showed peaks for both oxygen and carbon (Fig. 7a) but no distinguishable peak for sulphur, phosphorus or any other element. Comparison of spectra from the resin and formvar outside the root segment (Fig. 7b) showed a drop in the level for carbon and no change for sulphur or phosphorus.
The anatomy of the hair roots of W. pungens is consistent with previous descriptions for hair roots in the Ericaeae and Epacridaceae. The root surface is covered with a mucilaginous sheath as seen in other members of these two families (Leister, 1968; Bonfante-Fasolo & Gianinazzi-Pearson, 1979; Peretto et al., 1990; Ashford et al., 1996). The outermost cell layer of the hair roots is shown to be a true epidermis as in other species examined (cf. Peterson et al., 1980; Ashford et al., 1996; Read, 1996; Steinke et al., 1996), although the outer layer had in some older literature been identified as cortical tissue. The underlying cortical layer is identified as an exodermis, confirmed by the layer of suberin in the wall, and the endodermis by the well developed Casparian strips. A similar differentiation of the cortex was found in Lysinema ciliatum (Ashford et al., 1996) and Dracophyllumsecundum although in the latter, endodermal cells were collapsed so that the cortex appeared to be single-layered (Allen et al., 1989).
The thick multilamellate wall region of W. pungens epidermal cells with its distinctive pattern of staining is typical of the helicoidal-type of secondary wall described by Roland et al. (1987, 1993) and found in many cell types from a variety of plants and fungi (Emons et al., 1992; Wolters-Arts et al., 1993; Emons, 1994; Miller & Jeffries, 1994; Reis et al., 1994). Helicoidal microfibril arrangements also occur in primary walls but they are not multilamellate (Wolters-Arts et al., 1993). The data indicate that the thick wall laid down in W. pungens epidermal cells is a secondary wall with a typical multilamellate arrangement of microfibrils. Epidermal cells of Calluna vulgaris L. (Ericales) hair roots are also reported to develop helicoidal walls (Peretto et al., 1990; Perotto et al., 1995). The common denominator for the differentiation of a helicoidal-type secondary wall appears to be cell size and shape; cells with diameters between 15 and 60 µm lay down a helicoidal pattern (Emons & Kieft, 1994). The diameters of the epidermal cells in W. pungens are within this range.
Normally, a cellulosic wall microfibril pattern is only clearly revealed after extraction of wall matrix material (Reis et al., 1994; Emons & Mulder, 1998) but may be rendered visible here by extraction of water soluble matrix components during the long fixation period (see McCleary et al., 1976; Reis et al., 1994). Alternatively, it is possible that polysaccharides associated with cellulose microfibrils are staining (Emons et al., 1992). The spongy wall regions in W. pungens resemble the spongy thick-wall laid down in hair root epidermal cells of Lysinema ciliatum in the absence of mycorrhizal colonisation (Ashford et al., 1996). Spongy wall has so far only been found in colonised cells of W. pungens, but it is not possible to say whether it is induced by fungal colonisation or reflects the greater longevity of colonised cells. It is also not clear whether additional wall layers have been laid down as spongy wall, or whether this is a modified helicoidal wall.
The various staining reactions provide an insight into some of the types of molecules present in hair root epidermal cell walls. The reaction with Calcofluor White M2R, in the absence of staining with aniline blue, indicates β-1,4 D-glucosyl residues, consistent with the presence of cellulose, rather than mixed linkage β-glucans (Hughes & McCully, 1975; Wood & Fulcher, 1978; Wood et al., 1983; Evans et al., 1984). Bonfante-Fasolo et al. (1990) found heavy labelling by cellobiohydrolase-gold over the helicoidal wall regions of Calluna vulgaris hair root epidermal cells confirming cellulose in this species. The positive PAS reaction indicates 1 : 2 glycol groups, as found in hexose sugars such as glucose, mannose and galactose (Pearse, 1968). The strong positive reaction by peanut agglutinin and Bandeiraea simplicifolia I lectin indicates abundant galactose side chains, whilst the negative reaction from ConA indicates that glucose or mannose side chains were negligible, or not available for staining. Staining with both toluidine blue and alcian blue 8GX indicates that acidic groups are present throughout the thick wall. The intense metachromasy with toluidine blue at pH 4.4 but not at pH 1.0 suggests that these are carboxyl rather than sulphate. The progressive loss of staining as MgCl2 (≥ 0.5 M) is added to alcian blue 8 Gx staining solutions at pH 5.7, also indicates carboxyl groups (Pearse, 1968). This result was consistent with the X-ray analysis where no sulphur was detected in spot analyses of the thick wall.
Although there appear to be significant amounts of carboxyl groups throughout the walls, staining by the two JIM antibodies was low. Lack of reaction with JIM5, indicates no/very low amounts of un-esterified pectin (polygalacturonic acid) or pectin with up to 50% esterification. The low level of binding to JIM7 indicates low levels of esterified pectin as well (Knox et al., 1990), at least with the JIM7 epitope. Similar results were obtained in Calluna vulgaris hair root epidermal cells (Peretto et al., 1990; Perotto et al., 1995). This contrasts with the strong binding to the cortical and stelar cell walls, indicating relatively high amounts of unesterified and esterified pectins, a difference also found in roots of members of Gramineae and Chenopodiaceae (Knox et al., 1990), as well as C. vulgaris hair roots (Peretto et al., 1990) and mycorrhizal roots of Allium porrum cv Mastro di Carenta (Bonfante-Fasolo et al., 1990). A range of other species, including carrot roots do not show this difference between epidermal and other root cells (Perotto et al., 1997).
Perotto et al. (1995) consider that the low levels of pectin are significant in endophyte infection. While some C. vulgaris endophytes secrete polygalacturonases into the medium during the saprotrophic phase, these enzymes are not detected on the surface of hyphae in the epidermal cells of host roots. They have been detected on hyphae on the surface of the host roots, and on hyphae inside nonhost roots where the walls are demonstrated to be rich in pectins. This has led to the suggestion that absence or very low amounts of pectins in the epidermal cell walls in C. vulgaris may cause the ericoid fungus to ‘switch’ from a saprophytic to a less potentially damaging mycorrhizal phase where pectinase enzymes are turned off (Perotto et al., 1995). The data for W.pungens agree with the findings for C. vulgaris, suggesting that this phenomenon may be more widespread.
Large numbers of galactose-rich side chains are present in wall polymers such as galactomannans and arabinogalactan-proteins. It is unlikely that there are large amounts of arabinogalactan-proteins in the thick wall in view of the lack of a positive reaction with general protein stains such as amido black 10B, although they may have been lost in processing due to their high water solubility (Fry, 1994) or masked in some way. Galactomannans are also water soluble, depending on the degree of galactose substitution and may similarly be lost during processing unless cross-linked to other wall components (McCleary et al., 1976; Matheson, 1984). Such hydrophilic wall components may have a role in the water relations of the epidermal cells. W. pungens grows in a sandy heath-type habitat subject to frequent periods of drying out and re-wetting. Inclusion of galactose-rich, hydrophilic polysaccharides, such as galactomannan, in the thick wall may buffer the epidermal cells (and their hyphal coils) from transient changes in soil water status, promote survival of sloughed epidermal cells, and influence outgrowth of mycorrhizal hyphae. Polymers such as galactomannan may also act as a substrate for the mycorrhizal fungus. There is a parallel in the endosperm of some seeds especially those of legumes where galactomannans with a galactose content ranging from 15 to 50% are stored apparently as unusually thickened secondary walls (Matheson, 1984). Galactomannans are mobilized during germination and act as storage reserves (Reid, 1971; McCleary et al., 1976; Ashford & Gubler, 1984). However, their primary role is thought to be in maintaining a moist environment for the germinating embryo during drying conditions, as an adaptation to semiarid environments (Reid & Bewley, 1979), and this may also be the case in epacrid hair roots.
The JIM5 and JIM7 antibodies were kindly provided by the Monoclonal Laboratory, John Innes Centre, Norwich, UK. Field Emission Scanning Electron Microscope data was collected with the help of Viera Piegerova, Electron Microscope Unit UNSW. Thanks are also due to Louise Cole and Danielle Davies for technical assistance. We thank the Australian Research Council for the grant which supported this project.