Cell wall alterations and localized accumulation of feruloyl-3′-methoxytyramine in onion epidermis at sites of attempted penetration by Botrytis allii are associated with actin polarisation, peroxidase activity and suppression of flavonoid biosynthesis

Authors

  • Sarah R. McLusky,

    1. Department of Biological Sciences, Wye College, University of London, Ashford, Kent TN25 5AH, UK, and
    2. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author
  • Mark H. Bennett,

    1. Department of Biological Sciences, Wye College, University of London, Ashford, Kent TN25 5AH, UK, and
    2. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author
  • Michael H. Beale,

    1. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author
  • Mervyn J. Lewis,

    1. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author
  • Paul Gaskin,

    1. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author
  • John W. Mansfield

    1. Department of Biological Sciences, Wye College, University of London, Ashford, Kent TN25 5AH, UK, and
    2. I.A.C.R., Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol BS41 9AF, UK
    Search for more papers by this author

  • The Plant Journal (1999) 17(5), 523–534

*For correspondence (fax +12 338 13140; e-mail j.mansfield@wye.ac.uk ).
†Joint first authors.

Summary

Granular deposits of reaction material (RM) were formed in onion epidermal cells at sites of attempted penetration byBotrytis allii. Both RM and the adjacent cell wall fluoresced blue under UV excitation. The blue autofluorescence was caused primarily by the accumulation of feruloyl-3′-methoxytyramine (FMT) and feruloyltyramine (FT) within challenged tissues. Additional phenolics increasing at infection sites were identified as coumaroyl glucose, coumaroyltyramine (CT) and 2-hydroxy-2-(4-hydroxyphenyl) ethylferulate. The major autofluorescent components of RM, the hydroxycinnamic acid amides FMT and FT, were bound by ether linkage onto the cell wall as well as being present in methanol soluble granules. Formation of RM was associated with early increases in peroxidase activity detected by histochemistry at reaction sites and striking polarisation of actin microfilaments. Quantitative analysis of quercetin and cyanidin glucosides revealed that the localized synthesis and deposition of feruloyltyramine derivatives was associated with suppression of flavonoid and anthocyanin accumulation in a zone of cells around those accumulating RM. No antifungal activity was detected in FMT, FT or CT, nevertheless it is proposed that the phenolics have a key role in resistance by preventing fungal degradation of the cell wall.

Abbreviations
CC

column chromatography;

CT

coumaroyltyramine;

DAD

diodearray detector;

ESMS

electron spray mass spectrum;

FT

feruloyltyramine;

FMT

feruloyl-3′-methoxytyramine;

HCA

hydroxycinnamic acid amide;

HHPEF

2-hydroxy-2-(4-hydroxyphenyl) ethylferulate;

RM

reaction material;

TMS

trimethylsilyl.

Introduction

The highly localized alteration of the plant cell wall and formation of paramural deposits (papillae) at sites of attempted penetration by fungal pathogens have been described as important forms of plant defence (for reviews see Aist 1976; Aist & Gold 1987; Mansfield 1990; Nicholson & Hammerschmidt 1992; Ride 1983, 1986). Modification of the plant cell wall was recognised as a potential mechanism of resistance by Young (1926), in one of the first reports on the reaction of plants to fungal challenge. He described the response of 78 species and varieties of plants to inoculation with Alternaria spp. and other leaf-spotting fungi and found that failure to penetrate was commonly associated with ‘thickenings’ in the cell wall and formation of local additions or ‘callosities’. He showed that the structures of deposits formed at reaction sites were strikingly different between plant species. For example, he illustrated the formation of smooth hemispherical callosities in wheat and barley, whereas in onion and sorghum reactions involved the accumulation of granular deposits which in the latter case developed a deep red/brown coloration.

Subsequent research has shown that although the appearance of reaction sites may vary between species, a common biochemical feature of cell wall alteration is the accumulation of phenolic compounds both in the wall itself and the underlying deposits ( Bestwick et al. 1995; Matern & Grimmig 1994; Nicholson & Hammerschmidt 1992). Phenolics are thought to be the cause of autofluorescence observed at reaction sites examined by UV epifluorescence microscopy ( Bennett et al. 1996; O’Neill & Mansfield 1982). In sorghum, granules aggregating at penetration sites have been found to contain the deoxyanthocyanidin phytoalexins apigeninidin and luteolinidin, and therefore present an efficiently localized fungitoxic barrier to fungal invasion ( Snyder & Nicholson 1990; Snyder et al. 1991).

The mechanisms of site-directed secretion of phenolics into papillae and their incorporation into the cell wall are poorly understood but they represent excellent examples of subcellular co-ordination occurring within plant cells, as discussed by Bolwell (1993) and Brown et al. (1998). Major cytoskeletal rearrangements have been observed following fungal challenge in barley, parsley and potato ( Freytag et al. 1994; Gross et al. 1993). For example, both microtubules and actin microfilaments became orientated towards penetration sites in barley coleoptile cells challenged by the pea pathogen Erysiphe pisi ( Kobayashi et al. 1997). In bean leaves, deposition of phenolics in mesophyll cell walls in response to the soybean rust fungus is greatly reduced by anti-microfilament agents ( Perumalla & Heath 1991).

The response of onion bulb scale epidermal cells to attempted penetration by species of Botrytis was examined by Stewart & Mansfield (1984, 1985). Confirming the earlier observation by Young (1926), they found that granular deposits, described as reaction material (RM), accumulated to line the inner face of the cell wall outside the plasma membrane. The response was particularly striking following inoculation with low concentrations of Botrytis allii conidia. Microautoradiography, using 3H and 14C labelled phenylalanine and cinnamic acid as precursors indicated the synthesis of phenolic material in the endoplasmic reticulum and the accumulation of label within granules and the overlying cell wall. Formation of reaction material was associated with the inhibition of fungal growth within the modified wall ( Stewart & Mansfield 1985).

In this paper we describe four new aspects of the response observed in onion cells: (i) the identification of feruloyl-3′-methoxytyramine as a major component of the granular deposits; (ii) the finding that deposition of hydroxycinnammic acid amides (HCAs) is associated with suppression of flavonoid and anthocyanin accumulation; (iii) the detection of peroxidase activity at reaction sites; and (iv) the role of actin microfilaments in directing secretion and also in the intercellular signalling which co-ordinates the overall response. The finding that feruloyl-3′-methoxytyramine and related derivatives accumulate in a free form in onion provides a marked contrast with parsley and potato cells in which HCAs are reported to be bound rapidly into plant cell walls during the expression of resistance to fungal infection ( Clarke 1982; Keller et al. 1996).

Results

Macroscopic symptoms and microscopical observations

Following dissection of fleshy scales from red onion bulbs, their inner epidermis accumulated red pigments within 3 days. However, at sites of fungal inoculation the tissue remained white, the lack of pigmentation extending slightly beyond the area covered by inoculum droplets, as illustrated in Fig. 1. Microscopy revealed that pigments were located in cell vacuoles and were usually not produced in tissue challenged by B. allii unless at least three or four cells distant from a site of attempted penetration. The transition between white to pigmented tissue was fairly abrupt, with only a limited gradation in colouring observed ( Fig. 1). Pigmentation was sometimes enhanced around the edge of the white zone.

Figure 1.

Suppression of anthocyanin accumulation at inoculation sites.

(a) Fleshy bulb scales. From left to right, a freshly prepared segment, tissue inoculated with water alone and a scale inoculated with a droplet of a suspension of B. allii conidia for 3 days; 0.5× life size.

(b) Epidermal strip from an inoculation site showing the interface between red and white tissue. The tissue was plasmolysed in 1 m KNO3. Note the deposition of reaction material at a site of attempted penetration (arrowed) several cells away from the zone of pigment accumulation; bar, 200 μm.

Granular deposits of RM accumulated at sites of attempted penetration by B. allii; the granules fluoresced pale blue under UV radiation ( Figs 1 and 2). Plasmolysis of responding cells revealed that the main granular deposits were located between the plasma membrane and cell wall, although small fluorescing granules remained within the cytoplasm in plasmolysed cells. Soaking tissue in methanol removed autofluorescent material from granules but blue autofluorescence remained within the epidermal cell wall around the invading infection hypha ( Fig. 2). Although blue fluorescence was removed from granules by organic solvents, a residual non-autofluorescent matrix often remained to define the site of RM deposition (not shown).

Figure 2.

Removal of autofluorescence from granular deposits of reaction material.

(a) Sites of attempted penetration (large arrows) 2 days after inoculation observed under tungsten light illumination: granular reaction material has accumulated at both sites, strands of cytoplasm (small arrows) radiate from deposits.

(b) Granules and altered cell walls fluoresce blue under UV excitation.

(c) Flooding the preparation with methanol removes autofluorescent granules but fluorescence is retained by the cell wall (UV excitation); bar, 100 μm.

Time course studies revealed that cytoplasmic streaming directed towards the site of attempted penetration was the first visible response to challenge 6–8 h after inoculation, during the formation of appressoria by B. allii. Blue autofluorescence first appeared in the cell wall about 10 h after inoculation. Obvious aggregates of autofluorescent granules were not observed until 2–3 h after the initial response but subsequently increased in number and size ( Fig. 2) until about 3 days after inoculation.

Identification of phenolic components of reaction material

Analysis of methanolic extracts of responding epidermal tissue by TLC demonstrated accumulation of one major and several minor blue fluorescing components which were absent from uninfected tissue. Because the amounts of tissue and induced phenolics that could be recovered using epidermal strips was limited, an alternative system was used to generate sufficient material for purification and structural characterization. Stabbing onion bulb tissue with a cocktail stick carrying Pseudomonas cichorii was found to produce large necrotic lesions surrounded by numerous cells accumulating granular autofluorescent deposits. Fractionation of extracts allowed the recovery of fluorescing compounds corresponding to those recovered from epidermal strips challenged by B. allii. The autofluorescent phenolics recovered were identified by 1H and 13C NMR, HPLC-ESMS, and also GCMS of their TMS derivatives. The ability of less polar solvents, such as CHCl3, to remove the autofluorescence from granules was correlated in extracts with the presence of the major blue fluorescent compound which was identified as feruloyl-3′-methoxytyramine* (FMT).

Additional blue fluorescing compounds associated with infected tissue, but present at lower levels than FMT, were identified as feruloyl tyramine (FT), coumaroyltyramine* (CT), feruloyl glucose and 2-hydroxy-2-(4-hydroxyphenyl) ethylferulate* (HHPEF). Structures of FMT, FT and CT ( Fig. 3) were confirmed by synthesis. The predominant derivative induced in epidermal strips was FMT. The autofluorescence of pure FMT or FT and that observed in granular RM could not be distinguished by fluorescence microscopy. Of the other phenolics, feruloyl glucose had a similar autofluorescence, whereas CT was darker blue and HHPEF purple/blue in colour. Results therefore indicated that FMT was the major autofluorescent component of granular RM.

Figure 3.

Autofluorescent phenolics identified in tissue containing granules of reaction material.

(a) Structure of hydroxycinnamic acid amides: coumaroyltyramine R1 and R2 = H; feruloyltyramine, R1 = MeO, R2 = H; feruloyl-3′-methoxytyramine, R1 and R2 = MeO.

(b) 2-hydroxy-2-(4-hydroxyphenyl) ethylferulate (HHPEF).

The localized accumulation of FMT and related derivatives represented formation of a potentially antimicrobial barrier to fungal invasion. The antifungal activities of FMT, FT and CT were therefore tested against conidia and germ-tubes of B. allii and B. cinerea, using the sensitive methods described by Rossall et al. (1980 ). The HCAs were all inactive at the maximum concentration tested, 400 μg ml–1 or approximately 10–3 M.

Characterisation of flavonoids and anthocyanins

In order to examine the accumulation of red pigments and associated phenolics in detail, the compounds accumulating within red tissues were also identified by 1H and 13C NMR and HPLC-ESMS. The major flavonoids recovered were quercetin and its glucosides including 4′ 3; 3, 4′ and 3,4′,7–0-glucoside ( Fossen et al. 1998 ). The major anthocyanins were identified by ESMS of isolated peaks from HPLC, as cyanidin-3-glucoside, cyanidin-3-acetylmalonyl glucose* and cyanidin-3-malonyl glucoside.

Analysis of changes in free HCAs, flavonoids and anthocyanins

HPLC systems were developed for quantitative analysis of phenolics within the onion epidermis. The solvent gradient used for flavonoids and anthocyanins was not ideal for FMT derivatives, therefore a second isocratic system was specifically developed to examine FMT accumulation. Chromatographs illustrating the resolution achieved are given in Figs 4 and 5, which also show the results of time course experiments on the accumulation of compounds in epidermis at inoculation sites and in surrounding tissue. Data are presented for total anthocyanins and flavonoids as there were few differences between the rates of accumulation of members of these classes of compound in reddening tissue. Accumulation of flavonoids mirrored the pattern observed with anthocyanins, both groups of compounds being present in low levels at inoculation sites. In contrast to the flavonoids, anthocyanins and coumaroyl glucose, the levels of feruloyl glucose increased both within and around the inoculation sites ( Fig. 4). FMT and related derivatives were only detected within inoculated tissues and increased in concentration as granules of RM were deposited ( Fig. 5).

Figure 4.

Accumulation of anthocyanins and flavonoids.

(a) Analysis by reversed phase HPLC of anthocyanins and flavonoids. Gradient elution of extracts of uninoculated epidermal tissue after 4 days incubation resolves anthocyanins and flavonoids (detection at 280 nm): CG = coumaroyl glucose, FG = feruloyl glucose, Q = quercetin, Q1 = quercetin 3,4′,7-triglucoside, Q2 = quercetin 3,4′-diglucoside, Q3 = quercetin 3-glucoside, Q4 = quercetin 4′-glucoside, A1 = cyanidin 3-glucoside, A2 = cyanidin-3-malonylglucose, A3 = cyanidin-3-acetyl- malonylglucose. Changes in the concentration of (b) flavonoids (solid lines) and anthocyanins (dashed lines); (c) coumaroyl glucose and (d) feruloyl glucose, within (•) and away (□) from sites inoculated with B.allii spore suspension, or water (X) as a control. Data in (b) represent the sum of all peaks identified as either flavonoid or anthocyanin (mean from three experiments). Flavonoids were quantified as quercetin, anthocyanins as cyanidin 3-glucoside. Data in (c) and (d) are the means of three replicates at each time point; bars ± SEM. Coumaroyl and feruloyl glucose were quantified as coumaric acid.

Figure 5.

Accumulation of autofluorescent phenolics.

(a) Isocratic elution was used to quantify hydroycinnamic acid amides at inoculation sites (detection at 320 nm), this output was from a sample collected 3 days after inoculation: CT = coumaroyltyramine, FT = feruloyltyramine, FMT = feruloyl-3′-methoxytyramine, HHPEF = 2 hydroxy-2-(4-hydroxyphenyl) ethylferulate.

(b) Changes in the concentrations of hydroxycinnamic acid amides and 2-hydroxy-2-(4-hydroxyphenyl) ethylferulate in onion epidermis challenged by B.allii. Data are the means of six replicates, quantified as FMT; bars ± SEM. Coumaroyltyramine (□), feruloyltyramine (▪), feruloyl-3′-methoxytyramine (♦), HHPEF (⋄), all compounds in water inoculated tissue (×).

Analysis of wall-bound phenolics

Although extraction with methanol removed autofluorescence from RM, residual fluorescence remained in the cell wall itself ( Fig. 2). The nature of the wall-bound phenolics at reaction sites was therefore examined.

Epidermal strips were recovered from inoculation sites after incubation for 4 days. Sites were chosen for the high frequency of occurrence of RM at sites of attempted penetration by B. allii. Tissue inoculated with water alone served as a control. Sequential analysis involved methanol extraction, alkaline saponification of residual walls (1 m NaOH at 70°C) and finally hydrolysis in 6 m NaOH by microwave bomb digestion. Treatment of pure FMT and FT by saponification caused little degradation; no loss of FT was observed and 5% of FMT was hydrolysed to vanillin and ferulic acid (comprising about 1 and 4% of the initial FMT). The microwave bomb hydrolysed FMT (100%) and FT (64%) to ferulic acid and vanillin and ferulic acid, vanillin and p-hydroxybenzaldehyde, respectively. Similarly severe treatment of ferulic acid caused 17% conversion to vanillin.

Saponification treatment did not significantly alter the autofluorescence observed in cell walls at reaction sites, but traces of p-hydroxybenzaldehyde and vanillin were released from control tissues and, in greater quantities, from infection sites, as summarized for a typical experiment in Table 1. Digestion in the microwave bomb caused the release of further amounts of aldehydes, and also ferulic and coumaric acids but only from infected tissue. It proved impossible to examine digested cell walls microscopically for autofluorescence after hydrolysis under severe conditions. As only traces of FMT and FT per se were released from cell walls by saponification, results suggest that the HCAs were not esterified onto the plant cell wall. Therefore, the vanillin, p-hydroxybenzaldehyde and ferulic acid released by microwave bomb digestion were probably derived from FMT and FT bound to cell wall components by ether linkages.

Table 1.  Release of phenolics from onion epidermis by sequential extraction with methanol, saponification of residual cell walls with 1 m NaOH and severe hydrolysis with 6 m NaOH in a microwave bomb
Yield (μg g–1 fwt) after extraction a
MethanolSaponificationBomb
 Control bWater Control cChallenged cControlWater ControlChallengedControlWater ControlChallenged
  • a

    If none detected, no data given.

  • b

    Yield from freshly prepared epidermis.

  • c

    Yield from epidermis inoculated with water (Water Control) or B.allii (Challenged) and incubated for 4 days.

Coumaroyl glucose 312.136.8      
Feruloyl glucose 10.527.3      
p-Hydroxybenzaldehyde    0.20.21.70.10.15.1
Vanillin   trace0.21.60.2trace5.8
Coumaric acid    trace0.6 trace3.2
Ferulic acid    trace0.7  3.5
Coumaroyl tyramine  5.7  0.3   
Feruloyl tyramine  12.8  0.6   
Feruloyl methoxytyramine  87.1  0.5   
HHPEF  22.3      

Localization of peroxidase and actin polymerisation at reaction sites

The apparent cross-linking of phenolics onto onion cell walls indicated involvement of peroxidase in the early responses observed. In order to examine peroxidase activity at reaction sites, histochemical staining with diaminobenzidine (DAB) as the substrate was utilized (DAB proved more effective than the alternatives, guiacol or tetramethylbenzidine). Localized brown staining, indicating peroxidase activity, was detected in the plant cell wall at reaction sites within 12 h of inoculation; it did not appear to extend into paramural granules of RM ( Fig. 6). Most sites with positive staining for peroxidase also displayed some autofluorescence from the responding cell wall. In some cases, during the early stages of the plant’s response, staining was detected in the absence of autofluorescence which suggested that a localized increase in peroxidase activity was the primary response.

Figure 6.

Localization of peroxidase activity by histochemical staining with diaminobenzidine, activity is indicated by the formation of red/brown deposits; bar, 100 μm.

(a) Cells 9 h after inoculation. Note the staining around the site of attempted penetration (arrowed); brightfield illumination.

(b) Cells 20 h after inoculation in which staining has extended down into the anti-clinal cell walls at two sites of penetration (arrowed); Nomarski optics.

The active streaming of cytoplasm towards sites of RM deposition strongly implicated a role for the cytoskeleton in co-ordinating the defence response in onion. Actin was therefore localized using a rhodamine-linked phalloidin probe for confocal microscopy. In unchallenged cells, fine actin filaments were observed running mainly in the long axis of the cell, with some radiating from the nucleus. Changes in this pattern occurred during the early stages of attempted penetration and were often associated with branching of filaments to form a network directed to the fungal challenge ( Fig. 7). Although our techniques did not allow the examination of peroxidase activity and actin filaments in the same cells, time course studies indicated that increases in enzyme activity and rearrangement of the cytoskeleton occurred at the same time after fungal challenge. Analysis of established deposits 1 day after inoculation revealed the striking polarisation of actin to form filaments which radiated from sites of RM deposition and sometimes extended from within the challenged cell into surrounding tissue. Excitation with the laser did not cause autofluorescence, but granules of RM were detected by their strong reflectance and at some sites were closely associated with actin filaments ( Fig. 7). Actin polarisation did not extend into all cells which failed to accumulate red pigments and therefore was not structurally required for suppression of flavonoid accumulation ( Fig. 1).

Figure 7.

Actin filaments located by rhodamine-phalloidin labelling and confocal microscopy in onion epidermis; bars, 100 μm.

(a) Control tissue 6 h after inoculation with water alone.

(b and c) Tissue 12 and 24 h after inoculation with Botrytis allii, repectively. In (a) note fluorescent actin filaments radiating from nuclei (arrowed). Fine filaments are present in all cells. Polarisation of filaments towards the fungus is clearly visible 12 h after inoculation, as shown in (b). Note the spore and germ-tube (arrowed) on the surface lying over the plant cell wall. Polymerized actin forms a network of filaments focusing towards the penetration site. In (c) a brightly fluorescent site of reaction material deposition is the centre of actin polarisation in surrounding cells. Note that autofluorescent granules of reaction material (arrowed) appear to be closely associated with filaments.

Discussion

Analysis of onion epidermis containing granular reaction material revealed, for the first time, not only the accumulation of FMT and HHPEF in plants challenged by fungi, but also the accumulation of much higher concentrations of free HCAs than previously reported. The autofluorescent granules observed at reaction sites were found to be particularly rich in FMT. Granules of RM observed within the cytoplasm were reported by Stewart & Mansfield (1985) to be bounded by a membrane. Fusion of the bounding membrane with the plasmalemma was thought to result in deposition and accumulation of phenolic deposits in the apoplast. The packaging of FMT and possibly other HCAs and HHPEF into the secretory granules now provides a useful model for the study of exocytosis in plant cells, as discussed by Battey et al. (1999).

Increases in HCAs in response to fungal attack was first reported by Clarke (1982) in his analysis of Phytophthora infestans infected potato. Since then, their stress-induced synthesis has been reported in many plants, particularly members of the Solanaceae ( Negrel et al. 1996; Pearce et al. 1998). Increases in soluble and wall-bound HCAs have been observed in elicitor-treated cell suspension cultures of potato ( Keller et al. 1996; Schmidt et al. 1998), tobacco ( Negrel & Javelle 1995) and Solanum khasianum ( Mühlenbeck et al. 1996). The rapid and co-ordinated activation of the HCA biosynthesis pathway in response to elicitation, microbial challenge or wounding supports the idea that these compounds are produced as part of an active defence response ( Hohlfeld et al. 1996). Feruloyltyramine, but not feruloyl-3′-methoxytyramine, has been reported to accumulate in onion roots colonised by endomycorrhizal fungi ( Grandmaison et al. 1993).

The probable link between feruloyltyramine and flavonoid biosynthesis is outlined in Fig. 8. In onion, synthesis of HCAs appears to suppress accumulation of flavonoids and anthocyanins. Similar redirection of phenylpropanoid metabolism is also found in parsley, in which flavonoid synthesis is diverted to production of furanocoumarin phytoalexins following treatment with elicitors ( Hahlbrock et al. 1995; Lozoya et al. 1991). In sorghum, light-induced anthocyanin accumulation is also reduced in a zone of cells surrounding sites of deoxyanthocyanidin phytoalexin production ( Lo & Nicholson 1998), a pattern of plant response very similar to that observed in onion. The spatial regulation of the response observed in onion is intriguing. The signal(s) released which establish a zone designated to be free of flavonoids extend beyond the cells accumulating HCAs which only occur immediately surrounding the penetration site (see Fig. 1). The absence of flavonoid and anthocyanin accumulation does not therefore appear to require accumulation of the HCAs within the affected cell. Such a pattern of suppression cannot be explained simply by diversion of metabolites and is more likely to be associated with the repression of transcription of genes involved in the flavonoid pathway.

Figure 8.

Hydroxycinnamic acid amides, and flavonoids and anthocyanins are synthesized via different branches of the phenylpropanoid pathway.

Enzymes: PAL = phenylalanine ammonia-lyase, CHS = chalcone synthase, TyrDC = tyrosine decarboxylase, THT = tyramine hydroxycinnamoyl transferase.

Reaction material in sorghum is associated with the localized accumulation of the strongly antifungal deoxyanthocyanidin phytoalexins. However, we failed to detect significant antifungal activity against Botrytis in FMT, FT or CT, so their role in resistance in onion is, therefore, not to form a barrier of fungitoxicity. We have not investigated the possibility that the known phytoalexins from onion, the tsibulins ( Dmitriev et al. 1990), are also present in RM. We propose that free hydroxycinnamoyl amides are secreted into the onion apoplast to provide a pool of precursors for peroxidative cross-linking into the cell wall to produce a physical barrier to penetration. The altered wall at reaction sites has already been shown to be resistant to degradation by cell wall degrading enzymes ( Stewart & Mansfield 1985). Our experiments on the digestion of onion cell walls indicated ether linkage of ferulic acid amides to wall components. A similar conclusion was reached by Negrel et al. (1996) from their analysis of potato periderm. We have demonstrated the potential generation of phenolic aldehydes from the amides during hydrolysis of the cell wall. Aldehydes such as benzaldehyde, syringaldehyde and vanillin have been described as ‘cell wall phytoalexins’ by Kauss et al. (1993) and have been shown to be released by mild alkaline hydrolysis from walls of grasses, parsley, potato and lettuce ( Bennett et al. 1996; Hartley & Keen 1984; Keller et al. 1996). Methylation increased the sensitivity of HCAs to hydrolysis and aldehyde formation. Although such treatment did not cause major conversion of the amides tested here, more labile derivatives may prove a source of the corresponding aldehydes in some plants, e.g. an ether-linked feruloyldimethoxytyramine might yield syringaldehyde after saponification.

Cross linking HCAs into the onion cell wall would require peroxidase activity ( Iiyama et al. 1994; Negrel et al. 1996). Histochemical localization of peroxidase was based here on the oxidation of DAB which polymerizes rapidly on contact with the enzyme in the presence of H2O2 ( Bestwick et al. 1998; Frederick 1987). Thordal-Christensen et al. (1997) have used DAB, in the absence of added H2O2, to detect the active oxygen species within papillae formed in barley epidermis. We found no staining without the addition of H2O2 to onion epidermis, indicating that peroxide levels were limiting in onion. Localized staining attributed to peroxidase activity was observed in onion cell walls that were destined to accumulate autofluorescent phenolics. The increase in peroxidase in the cell wall may involve rapid synthesis and secretion of the enzyme, both of which may be regulated by local changes in pH and Ca2+ levels ( Bestwick et al. 1998; Brown et al. 1998; Sticher et al. 1981). Changes in Ca2+ levels may also help to explain the altered pattern of actin microfilaments observed during attempted penetration by B. allii. Local increases in Ca2+ have been reported to stimulate directed assembly of actin in plant cells ( Kropf & Quatrano 1987). Regulation of both peroxidase activity and actin polymerisation by common factors is indicated by their coincident and early detection following inoculation.

Cytoskeletal rearrangement in response to fungal ingress may be involved in the dispersal of signals for activation of biosynthetic activities in addition to providing a route for the transport of phenolic products to the penetration site. The spatial and temporal patterns of cellular responses observed in onion epidermis indicate the possible involvement of signals generated at the reaction site effecting three separate pulses of response: (i) a primary elicitation of cytoskeletal rearrangement and increase in peroxidase activity at the penetration site; (ii) activation of phenolic biosynthesis in the penetrated and nearby cells; and (iii) a more widespread suppression of flavonoid biosynthesis in cells distant from the penetration site. Whether or not the same signal produced from the initial interaction between B. allii and the onion cell wall elicits each part of the co-ordinated response or whether a primary elicitor activates a signalling cascade remains to be determined. Onion epidermis would now appear to be well suited for further studies on both intra- and intercellular controls of the exocytosis required for deposition of HCAs and their incorporation into the cell wall. Preliminary experiments indicate that a key component may be an elicitor released from the plant cell wall by enzymes secreted by B. allii during the early phase of attempted penetration (J. Tuohy, S. McLusky and J. Mansfield, unpublished data).

Experimental procedures

Plants and inoculum

Red onion bulbs (Allium cepa L.) were purchased from a local store. Single mid bulb scales were cut into pieces, 3–4 cm wide and placed in clear plastic boxes (18 × 12 × 5 cm) base lined with moist tissue paper, adaxial surface uppermost. The onion segments were incubated at 20°C in continuous light. Suspensions of conidia of Botrytis allii were prepared as described by Stewart & Mansfield (1984) and inoculated onto the exposed epidermal surface. For microscopy, 20 μl droplets of a suspension of 5 × 103 spores ml–1 were usually used. Larger droplets, 150 μl with 105 spores ml–1, were inoculated for biochemical analyses (as shown in Fig. 1) and after 24 h the droplets were removed. Droplet removal was found to increase the frequency of RM deposition in response to the higher concentration of spores.

Microscopy

Basic studies. Epidermal strips prepared from inoculum sites were mounted in water for observation of fungal development and plant responses. Slides were flooded with 1 m KNO3 to plasmolyse cells, and methanol and other solvents to dissolve granular reaction material ( Stewart & Mansfield 1985). Autofluorescence was examined using a Nikon Optiphot microscope using UV or blue excitation (filter sets UV-2 A and B-3 A, respectively) or a Zeiss Axioplan microscope with filter sets 01 and 09 (for UV and blue excitation) and fitted with a Minolta RD–175 digital camera. Images were processed using Adobe Photoshop 4.0.

Peroxidase detection. Enzyme activity was detected histochemically with three substrates; 3,3′-diaminobenzidine-HCl (DAB), guiacol and 3,3′,5,5′-tetramethylbenzidine (TMB) using the methods outlined by Thordal-Christensen et al. (1997 ), Maehly & Chance (1954) and Imberty et al. (1984 ), respectively. The DAB assay proved the most sensitive and was usually performed with 0.1 m m H2O2 added to the 1 mg ml–1 DAB solution used to incubate epidermal strips for 30 min in the dark at 22°C. Strips were mounted in water or 50% glycerol for microscopy. No staining was observed in the absence of H2O2, after boiling epidermal strips, or after the addition of the peroxidase inhibitor, sodium azide (2 m m).

Actin localization. Microfilaments were detected by labelling with phalloidin linked to rhodamine using modifications of the methods described by Wulf et al. (1979 ) and Kobayashi et al. (1997 ). Epidermal strips were soaked in microtubule stabilising buffer (MSB, containing 0.1 m Pipes at pH 6.9, 1 m m MgSO4, 2 m m EDTA, 1% formaldehyde) for 10–15min. Wells on a 10-well multi-test slide (Icon Chemicals) were filled with extraction buffer (30 μl) comprising 300 μm MBS (maleimido benzoic acid N-hydroxysuccinimide ester), 0.05% IgePal and 0.33 μm Rhodamine-phalloidin (Molecular Probes Inc. USA) in MSB. An epidermal strip was floated in each well and the slide was incubated in a foil-covered Petri dish for 30–60 min. The epidermal strips were washed by transfer to MSB-filled wells on a clean slide. The buffer was drawn off the tissue sections using filter paper and replaced with fresh buffer and a coverslip added which was sealed with nail varnish. Labelled tissues were kept in the dark as much as possible to minimize fading and viewed quickly using a Leica TCS 4D microscope fitted with an Omnichrom Ar–Kr laser and Leica SCANware for Windows software. Phalloidin-linked fluorescence was visualized specifically with a green laser (TRITC filter).

Chromatography and spectroscopy

Routine methods and standard equipment were used for HPLC, HPLC-DAD, TLC and CC and for the determination of UV spectra. 1H and 13C NMR spectra were recorded on a JEOL GX400 instrument using CD3OD (hydroxycinnamic acid derivatives) or (CD3)2SO (flavonoid glycosides) as solvents. GC–MS was carried out on a VG7070 mass spectrometer coupled to a HP5890 GC. Samples containing HCAs were derivatized as TMS ethers using N-methyl-N-trimethylsilyltrifluoroacetamide and analysed on a BP-X5 capillary column (25 m × 0.2 mm) eluted with He at 100 kPa and temperature programmed from 25°C to 200°C at 10°C min–1 and then at 4°C min–1–300°C. Electrospray MS and MS2–4 were carried out on a Finnigan LCQ instrument. Samples were introduced directly as solutions in acetonitrile-acetic acid (99:1) or methanol-acetic acid (99:1) or by means of a coupled HPLC using acetonitrile-1% aqueous acetic acid solvent mixtures at a flow rate of 0.5 ml min–1.

Purification of flavonoids

Non-inoculated segments of bulb scales were incubated for 7 days. The red adaxial epidermal tissue (67 g) was removed, homogenized in 3 × 600 ml of methanol:water (7:3) and filtered through Miracloth Bio 101 (Calbiochem, Nottingham, UK). The extracts were combined, reduced to about 200 ml in vacuo, defatted by extraction with hexane (2 × 200 ml) and then further reduced to 40 ml. The crude extract was fractionated by passage through Sephadex LH20 (90 g) with a linear gradient of A (99% water:1% formic acid) to B (99% methanol:1% formic acid) over 4 h followed by isocratic elution with B for a further 5 h at a flow rate of 2.5 ml min–1; fractions were analysed by TLC and HPLC. Five compounds (A–E in reverse order of elution) were isolated from pooled fractions either directly or following further chromatography. All compounds were characterized by UV, 1H and 13C NMR and HPLC-ESMS. In addition, glycosides were hydrolysed and the products co-chromatographed with authenticated standards. Compounds were recovered as follows: (A) The precipitate formed on storage at 4°C was dissolved in methanol (30°C) and re-precipitated to give quercetin (18.5 mg). (B) Treated as A to give quercetin 4′-O-glucoside (180 mg). (C) After reducing the volume of fractions in vacuo the sample was purified through Sephadex LH20 (10 g) eluting with methanol to give quercetin 3-O-glycoside (39 mg). (D) Treated as C, but eluting with methanol:water (1:1) followed by reverse phase column chromatography (RPCC) over Lichroprep RP18, 40–63 μm (5 g) from Merck, eluting with methanol:water (1:1) to give quercetin 3–4′-O-glucoside (44.4 mg). (E) Treated as D but eluting RPCC with water:methanol (3:1), quercetin 3–4′-7-O-glucoside (32.4 mg) was precipitated from the eluate by the addition of diethylether.

Isolation of compounds induced by fungal challenge

Preliminary analysis by TLC of methanolic extracts of onion epidermis infected with B. allii demonstrated the accumulation of several compounds which fluoresced blue under UV light. The same compounds were also present in onion segments infected with the bacterial pathogen Pseudomonas cichorii. Onion segments (600 g) were stab inoculated at multiple sites with a cocktail stick dipped into a suspension of 1×109 bacteria ml–1. After 3 days the tissue was homogenized in 3×1.5 L methanol, filtered, the extract reduced to 400 ml, water added to give a final volume of 1 l and acidified to pH 4 with HCl. The liquor was extracted with 3×300 ml of ethylacetate, the extract dried over anhydrous sodium sulphate and reduced in vacuo. The residue was chromatographed over 20 g of silica gel (40–63 BDH) eluting with a linear gradient of chloroform:acetone:acetic acid (90:9.5:0.5) to acetone acetic acid (99.5:0.5) in a total volume of 2 L. The components of individual fractions were visualized under UV following TLC, and the fractions containing blue fluorescent compounds pooled into three samples (A–C in order of elution) for further purification. In A, two major components present were separated by CC over silica (10 g) eluting with hexane:ethylacetate:acetic acid (66.25:33.25:0.5) followed by CC of individual compounds through Sephadex LH20 (10 g) eluting with methanol. One compound yellow/brown in colour was identified as quercetin (by UV and 1H NMR). The other compound (5.7 mg) which fluoresced purple/blue under UV was identified as 2-hydroxy-2-(4-hydroxyphenyl)-ethylferulate (HHPEF) by UV,1H,13C NMR, HPLC-ESMS and GCMS of the TMS derivative. In B, quercetin (5 mg) and two compounds which fluoresced bright blue under UV were separated as for A, with the exception that the initial solvent used was an equal mixture of hexane:ethylacetate containing 0.5% acetic acid. The two fluorescent compounds were identified as N-trans-feruloyltyramine (FT, 1.6 mg) and N-trans-feruloyl-3′-methoxytyramine (FMT, 4.0 mg) by UV,1H 13C NMR,HPLC-ESMS, and GCMS of the TMS derivatives. With pool C, CC twice over silica (5 g) eluting with ethlyacetate:acetic acid (99.5:0.5) gave a residue of 9.4 mg containing blue and purple fluorescing components which were further purified by semi-prep HPLC on Spherisorb ODS2 (10×250 mm) eluting with a gradient of 99%A (water+1%formic acid), 1%B (acetonitrile) to 65%A, 35%B over 30 min to give two compounds; trans-feruloyl glucose (1.2 mg) and trans-coumaroyl glucose (0.3 mg). The cinnamoyl glucosides were identified by UV,1H,13C NMR and HPLC-ESMS.

Identification of other compounds

Anthocyanins, aldehydes, coumaroyltyramine and ferulic acid were identified by co-migration with authenticated standards (TLC, HPLC-DAD) and by HPLC-ESMS.

Synthesis of N-trans-feruloyl-3′-methoxytyramine and other standards

The amide was formed by the condensation of ferulic acid and 3′-methoxytyramine in the presence of N,N′-dicyclohexylcarbiimide ( Tanaka et al. 1989 ). The following compounds were synthesized for comparative analyses using the same strategy: coumaroyl-, caffeoyl- and feruloyl-tyramine, methoxytyramine and octopamine; cinnamoyl- and sinapoyl-tyramine; and feruloyl-phenethylamine, norepinephrine, and dopamine.

Analysis of flavonoids, anthocyanins and cinnamoyl glucosides

Epidermal strips were peeled from onion bulb segments. A cork borer (10 mm diameter) was used to cut out the epidermis beneath the inoculum droplet and tissue around the site. Tissue at least 2 cm away from the inoculation site was also collected. Ten samples (approximately 50 mg) from each of the sites were soaked in 400 μl of 70% methanol for 16 h at room temperature, the tissue was then re-extracted with a further 100 μl of 70% methanol for 4 h. The extracts were combined and passed through a 0.45 μm filter. This method achieved recovery of > 95% of each of the major compounds. Separation was by reverse-phase HPLC at 30°C using a μBondapak, 10 μm 3.9×300 mm column (Waters) protected by a guard column containing Resolve C18 (Waters). Optimal separation was achieved using a 20 μl injection loop and a linear gradient over 30 min comprising an aqueous phase (2:2:96, acetic acid:phosphoric acid:water) and acetonitrile, with initial and final mixtures of 95:5 and 80:20 (aqueous:acetonitrile, respectively) and a flow rate of 1.5 ml min–1. Quantification was based on calibration curves for quercetin (flavonoids), cyanidin 3-glucoside (anthocyanins) and p-coumaric acid (cinnamoyl glucosides) at 360 nm, 280 nm and 320 nm, respectively, using Philips PU6000 software.

Analysis of HCAs and HHPEF

The extraction procedure was identical to that used for flavonoid analysis except that the extractions were carried out using 100% methanol. The compounds were separated as above but using a Spherisorb ODS2, 5 μm 10×250 mm column (Waters) and an isocratic solvent system of water:acetic acid:phosphoric acid:acetonitrile (67.2:1.4:1.4:30). Peaks were quantified from calibration curves of FMT measured at 320 nm.

Analysis of cell walls

Onion epidermis (about 0.5 g per sample) was extracted in 100% methanol, washed three times overnight in methanol and then dried at 40°C in vacuo. The residual tissue was saponified by resuspension in 3.6 ml of freshly prepared, degassed 1 m NaOH in a sealed glass vial, and heated to 70°C for 6 h. After cooling, 400 μl of acetic acid was added and the sample centrifuged at 14 000 g for 20min. The supernatant was then diluted with 5 ml of water, feruloyldopamine added as an internal standard and the phenolics released extracted with 3×5 ml ethylacetate. The ethylacetate was removed in vacuo, the residue resuspended in 1 ml MeOH and filtered (0.45 μm). The pellet remaining after centrifugation of the saponified extract was resuspended in 5 ml 4 m NaOH and transferred to a Teflon cup of a microwave bomb (Parr Instrument Co. 4780). This was heated at full power in a 750-W microwave oven for 90 sec; after cooling, 5 ml of water was added, the sample acidified with HCl and extracted as above. Both extracts were analysed by HPLC-DAD following separation on a Spherisorb ODS2, 5 μm, 4.6×250 mm column (Waters) eluting with a gradient of 90%A (water+1% acetic acid):10%B (acetonitrile) to 65%A:35%B over 50min, at 30°C and a flow rate of 1 ml min–1. Quantification was based on standard curves prepared for vanillin, p-hydroxybenzaldehyde (both at 280 nm) and ferulic acid (320 nm).

Acknowledgements

S.R.M. was funded by a research studentship from the BBSRC. Thanks are also due to Fiona McCann from the University of Kent, and Jon Green and Julie Dyhouse from Birmingham University, for advice on actin localization and use of their confocal microscopes. We must also acknowledge Mark. A. Bennett for production of the photographs.

Ancillary