Reduction of clubroot (Plasmodiophora brassicae) formation in Arabidopsis thaliana after treatment with prohexadione-calcium, an inhibitor of oxoglutaric acid-dependent dioxygenases

Authors


Abstract

Recently, flavonoids were shown to modulate the outcome of clubroot development in Arabidopsis thaliana after infection with the obligate biotrophic pathogen Plasmodiophora brassicae. Therefore, the development of clubroot disease was investigated in Arabidopsis after treatment with prohexadione-calcium (ProCa), an inhibitor of ascorbic acid/2-oxoglutaric acid-dependent dioxygenases such as flavanone-3-hydroxylase. The treatment resulted in a reduction of the flavonols quercetin and kaempferol in clubroots, whereas the precursor naringenin highly accumulated. The root system of ProCa-treated plants was better developed although galls were still visible. Thus, ProCa treatment resulted in reduced gall size. Flavonoids are thought to inhibit polar auxin transport by modulating auxin efflux carriers. It was investigated whether the auxin response might change as a consequence of the accumulation of naringenin in ProCa-treated plants. In the areas of gall development an auxin response was indicated by the auxin-responsive promoter DR5 coupled to the reporter β-glucuronidase (GUS), whereas very little staining was found in healthy root parts. No differences in GUS activity were found between P. brassicae-infected and ProCa-treated plants, and plants only infected with P. brassicae, indicating that the effect of ProCa treatment on clubroot reduction is not via changes in auxin responses. As ProCa is also an inhibitor of late steps in gibberellin biosynthesis, a specific gibberellin biosynthesis inhibitor, chlormequatchloride (CCC), was tested on club development. However, CCC did not reduce disease symptoms, indicating that the observed reduced gall development was not because of gibberellin biosynthesis inhibition by ProCa.

Introduction

The infection of cruciferous hosts with the obligate biotroph Plasmodiophora brassicae results in one of the most damaging diseases within this plant family. Plasmodiophora brassicae infects a range of economically important crop plants within the Brassicaceae, e.g. Brassica napus, B. oleracea, B. pekinensis and B. rapa, and can also infect the model plant Arabidopsis thaliana (Siemens et al., 2002). Plasmodiophora brassicae infection leads ultimately to cell elongation and cell division in infected roots and hypocotyls and results in the typical hypertrophied roots (clubroot; Ludwig-Müller et al., 2009). The life cycle of P. brassicae consists of the primary phase, which is restricted to root hairs and epidermal cells of the host, and the secondary phase, which occurs in the cortex and stele of roots as well as hypocotyls and leads to the abnormal development (Kageyama & Asano, 2009).

Plant growth hormones, especially cytokinins and auxins, which are both up-regulated during the secondary infection phase (Devos et al., 2005; Siemens et al., 2006), are involved in symptom development (for recent review see Ludwig-Müller & Schuller, 2008). A role for auxin during later stages might be specifically associated with the expansion of host cells. Consequently, modulation of auxin levels resulted in reduced clubroot symptoms (Ludwig-Müller et al., 1999). Auxin homeostasis can be controlled by biosynthesis, degradation or hydrolysis/synthesis of inactive conjugates, and transport. Also, secondary metabolites accumulate during clubroot development such as glucosinolates (reviewed in Ludwig-Müller, 2009) and flavonoids (Päsold et al., 2010). Flavonoids are thought to modulate auxin transport (Geisler & Murphy, 2006) and genetic and molecular evidence point to a role for flavonoids in gall formation by changing auxin transport properties, even though they are not essential compounds for gall formation (Päsold et al., 2010).

Flavonoids are a group of secondary products with an array of biological functions, including apparent roles in stress protection such as UV radiation (Winkel-Shirley, 2002). In addition, they are best known as the characteristic red, blue and purple anthocyanin pigments of plant tissues, which serve essential functions in plant reproduction by recruiting pollinators and seed dispersers (Winkel-Shirley, 2001). The phenylpropanoid pathway from which the flavonoids are derived links to plant defence (Ryder et al., 1987; Dixon, 2001). Some plant species synthesize 3-deoxyanthocyanins, which are involved both in defence (Snyder & Nicholson, 1990) and in pigmentation (Grotewold et al., 1994). Flavonoids with phytoalexin-like properties accumulate (Halbwirth et al., 2003; Spinelli et al., 2005; Rademacher et al., 2006) if plants are treated with prohexadione-calcium (ProCa), a plant growth regulator, which inhibits ascorbic acid and 2-oxoglutaric acid-dependent dioxygenases such as flavanone 3-hydroxylase (Fig. 1).

Figure 1.

Main steps in the biosynthesis of gibberellins and flavonoids indicating the enzymes which are inhibited by prohexadione-calcium (ProCa) and chlormequatchloride (CCC). Inhibition of oxoglutaric acid-dependent dioxygenases by ProCa is indicated by a cross and the target enzymes and the compound(s) accumulating are also shown in bold (coloured green online); inhibition by CCC is indicated with a cross and targets are also shown in bold (coloured red online). Metabolites shown in italics (coloured blue online) are those which could putatively be formed when eriodictyol is not converted to dihydroquercetin by flavanone-3-hydroxylase and a flavanone reductase is present. The position of the transparent testa6 (tt6) mutation is indicated. Dashed arrows indicate more than one reaction step, solid arrows indicate one enzymatic reaction. Compounds in boxes have been determined.

Being an inhibitor of several dioxygenases involved in gibberellin, ethylene and flavonoid synthesis, ProCa has an inhibitory effect on shoot growth via gibberellins, stimulatory effect on fruit set via ethylene inhibition and increased effect on biotic stress resistance via altered flavonoid patterns (Rademacher et al., 2006). The effect of ProCa application on the flavonoid patterns of several plant species, among them apples (Halbwirth et al., 2003; Spinelli et al., 2005; Rademacher et al., 2006) and grapevines (Puhl et al., 2008), has been shown. While an effect of ProCa as growth regulator, mainly in inhibiting gibberellin biosynthesis, is already well established (Rademacher, 2000), the influence on the patterns of flavonoids was only recently discovered (Puhl et al., 2008). If an unspecific flavanone reductase was present, treatment with ProCa resulted in the accumulation of the phytoalexin flavonoids eriodictyol and luteoliflavane (Puhl et al., 2008). Various diseases, caused both by microbes and insects, were shown to be reduced after application of the compound, probably due to the accumulation of the antimicrobial flavonoids (Bazzi et al., 2003; Halbwirth et al., 2003; Spinelli et al., 2005; Rademacher et al., 2006).

The reduction of diseases after ProCa treatment has been shown for several plant species, but not for Arabidopsis. As flavonoids are linked to altered club development in this species (Päsold et al., 2010), the objective of the current study was to evaluate the effect of ProCa on clubroot disease in P. brassicae-infected Arabidopsis. This work investigates whether the growth regulator ProCa has an effect on: (i) clubroot development, (ii) flavonoid accumulation, and (iii) auxin responsiveness, because flavonoids are thought to modulate auxin transport by inhibiting auxin efflux (Geisler & Murphy, 2006). In addition, oxoglutaric acid-dependent dioxygenases, which are inhibited by ProCa, are also involved in gibberellin (GA) biosynthesis, but there are other inhibitors of GA biosynthesis with different targets (Fig. 1). Chlormequatchloride (CCC) targets the steps from geranylgeranyl diphosphate via copalyldiphosphate to ent-kaurene catalysed by two enzymes, copalyldiphosphate synthase and ent-kaurene synthase (Rademacher, 2000). GAs have not been implicated in clubroot formation so far; a reduction after CCC treatment would suggest a role for this growth hormone as well. Therefore, this study also tested whether a GA synthesis inhibitor would result in the same effect on gall formation as ProCa.

Materials and methods

Plants and pathogens

The ecotype Col-0 of Arabidopsis thaliana was originally provided by Nottingham Arabidopsis Stock Centre (NASC). The line carrying the GUS gene under control of an auxin-responsive promoter (DR5::GUS) was kindly provided by Dr Tom Guilfoyle (University of Missouri, Columbia, USA). The P. brassicae isolate e3 used in this study was described by Fähling et al. (2003).

Cultivation and inoculation of plants

Cultivation and inoculation conditions of plants were performed according to Kobelt et al. (2000). Fourteen-day-old plants cultivated under a controlled environment (23 ± 1°C, 16 h light, 100 μmol photons s−1 m−2) were inoculated by injecting the soil around each plant with 2 mL of a resting spore suspension of the pathogen with a concentration of 106 spores mL−1. The spore suspension was obtained by homogenizing mature clubroot galls of B. rapa, followed by filtering through gauze (25 μm pore width) and two centrifugation steps (2500 g, 10 min).

Chemical treatment

Plants were treated with prohexadione-calcium (calcium 3-oxido-5-oxo-4-propionylcyclohex-3-enecarboxylate) using a soil drench method. As ProCa is sensitive to degradation by microorganisms (W. Rademacher, BASF, Limbugerhof, Germany, personal communication), a 25 or 100 ppm solution was applied every day over a period of 3 days to wildtype Arabidopsis and a 100 ppm solution to the DR5::GUS line. For the remaining 2 weeks, the plants were watered every second day with 250 mL of the respective solution. Treatment with chlormequatchloride (2-chloroethyltrimethylammonium chloride) with a 100 ppm solution was performed likewise.

Phytopathological analysis

Disease symptoms were assessed 28 days post-inoculation (dpi). For the quantitative estimation, the plants were cut at the top of the hypocotyl into shoots and roots. Forty control plants were compared to 40 inoculated plants for each treatment. Due to the small root size, five plants within each treatment were pooled to determine the root fresh weight. Thus, for the determination of means ± SE, eight replicates were included. To keep this comparable for shoot weight, the same five plants were also pooled and the fresh weight determined. Finally, the weight was calculated and presented per plant. The infected roots were measured as galls ignoring the remaining non-infected lateral roots. Root fresh weight was used to calculate a root index Ri/Rni (root fresh weight of infected plants divided by root fresh weight of non-infected plants) according to Ludwig-Müller et al. (1999) and Siemens et al. (2002). Likewise, an index for the shoot fresh weight (Si/Sni) was calculated. The length of the roots was determined for each plant and an index (Lri/Lrni) calculated as described above. The disease index (DI) was calculated by categorizing the individual roots into five classes (0 = no symptoms; 1 + 2 = roots with light symptoms; 3 + 4 = roots with severe symptoms) according to Siemens et al. (2002). Separate biological experiments were performed; each individual experiment consisted of a different number of Arabidopsis plants, with between 40 and 60 plants analysed for each treatment. The significance of the data was analysed first using the Kruskal–Wallis test and subsequently by comparing the mean rank differences as described by Siemens et al. (2002).

Determination of β-glucuronidase activity

The pattern of auxin distribution in IAA-responsive promoter-GUS lines (IAA2::GUS, DR5::GUS) was determined in roots by histochemical staining with 5-bromo-4-chloro-3-indolyl glucuronide (X-Gluc; Jefferson, 1987). Plants were incubated in 0·1 m NaPO3 buffer, pH 7·4, containing 10 mm Na2EDTA, 0·5 mm K3(Fe(CN)6), 0·5 mm K4(Fe(CN)6), 0·5% (w/v) Triton X-100 and 50 μm substrate (X-Gluc dissolved in DMSO). After 1 h incubation at 37°C, plants were rinsed and placed in 100% acetone for 30 min. After rinsing, the plants were transferred to the NaPO3 buffer (pH 7·4) overnight, without the substrate, to block the reaction.

HPLC determination of flavonoids

The method was similar to that described in Päsold et al. (2010). Briefly, flavonoids were extracted from c. 1 g fresh weight per sample in liquid nitrogen, then resuspended in 15 mL 70% methanol per gram fresh weight, extracted for 2 h at 8°C, centrifuged for 10 min at 13 000 g, and the supernatant evaporated to the aqueous phase. The latter was brought to pH 3 with 1 m HCl and then extracted twice with equal volumes of ethyl acetate. The organic fractions were pooled and evaporated to dryness. The residues were taken up in a final volume of 100 μL methanol for HPLC analysis. For the hydrolysis of flavonoid glycosides, the extract was hydrolysed with 1 part hydrolysis reagent (10 mL H2O, 100 mg BHT, 12 mL 2 m HCl) and two parts methanolic extract after centrifugation. The extract was incubated for 2 h at 85°C, with shaking every 15 min, then cooled on ice, centrifuged for 10 min at 13 000 g and brought to pH 3 with 1 m NaOH. The extract was then treated further as described for non-glycosylated flavonoids. The sample was completely injected onto HPLC (Jasco, Groß-Umstadt) using a 250 × 4 mm RP C18 column (Phenomenex) for separation. Solvents were 100% acetonitrile (A) and aqueous 2·5% acetic acid (B) and the flow rate 1 mL min−1. The gradient started with 3% A, followed by an increase to 9% A in 5 min, 16% A in 10 min, 50% A in 30 min, staying at 50% A for another 5 min before a washing (100% A) and an equilibrating step (starting conditions). Detection was performed with a multiwavelength photodiode array detector (Jasco) at 280 nm (naringenin) and 370 nm (kaempferol and quercetin). Calculations were performed on the basis of a standard curve with authentic standards as described in Päsold et al. (2010). The experiment was done in triplicate and repeated with independently cultivated plant material.

Transcript analysis

The AGI numbers of transcripts for flavonoid biosynthesis genes were located from the TAIR database (http://www.arabidopsis.org/). Transcript levels of flavonoid genes were compared for control and ProCa treatment by using the Arabidopsis eFP browser (http://bar.utoronto.ca; Winter et al., 2007).

Results

Reduction of clubroot symptoms in Arabidopsis with prohexadione-calcium

It was previously shown that flavonoids affect clubroot development most probably by modulating auxin transport and that flavonoid mutants were affected differently in their disease progression, but treatment with flavonoids did not result in a reduction of disease symptoms (Päsold et al., 2010; Table 1). Here, a different pharmacological approach has been used, based on the inhibition of flavanone 3-hydroxylase (F3H; Fig. 1) to increase or change endogenous flavonoid levels by ProCa treatment. To elucidate a possible mechanism of ProCa on clubroot disease development, a series of experiments were conducted testing: (i) disease severity; (ii) endogenous flavonoid contents; (iii) auxin response using plants harbouring an auxin-responsive promoter-reporter; and (iv) specificity by testing a growth regulator, chlormequatchloride (CCC) only involved in inhibiting gibberellin biosynthesis (Fig. 1).

Table 1. Disease indices (DI) after prohexadione-calcium and chlormequatchloride treatments in Arabidopsis thaliana infected with Plasmodiophora brassicae, in comparison to control infection and infection of the tt6 mutant, which results in naringenin accumulation. Inoculation was done with 107 spores mL−1
Plant/treatmentConcentrationDI control plantDI treated/mutant
  1. a

    Data for tt6 and Ler (corresponding wild type) taken from Päsold et al. (2010).

  2. b

    Naringenin treatment itself did not induce tolerance; the DI was calculated from the same experiment that is depicted for root and shoot indices in Päsold et al. (2010).

  3. c

    Significant differences at the  0·05 level are indicated by an asterisk.

Prohexadione-calcium25 ppm9048c
Prohexadione-calcium100 ppm9068c
Chlormequatchloride100 ppm9089
tt6 a 8263c
Naringeninb10 μm6561

Arabidopsis plants inoculated with P. brassicae were treated with ProCa and examined for disease development. The shoot parts of uninfected plants were smaller when treated with ProCa than the shoots treated with H2O only, as expected if GA biosynthesis is also inhibited (Fig. 2). In contrast, treatment of infected plants with ProCa led to greater shoot growth compared to the respective P. brassicae-infected water control. Treatment of P. brassicae-inoculated Arabidopsis plants with ProCa resulted in more vigorous roots, although galls were still present (Figs 3 & 6). To evaluate this further, different disease parameters were investigated, including the disease index (DI), the fresh weight index of infected to uninfected roots (Ri/Rni), and shoots (Si/Sni), as well as root lengths (Lri/Lrni). A small root index means less gall development and the low total root length index shows the reduction of the root system during infection. At 25 ppm ProCa, the root system including lateral roots was larger (Fig. 6). A large shoot index indicates more vigorous plants (Fig. 3). In addition the disease index (DI) as a measure of disease severity was calculated. For infected, untreated wildtype plants the DI was always above 80, which indicates high susceptibility, whereas for the ProCa-treated plants at both concentrations the DI was below 70 in all experiments (Table 1), indicating tolerance to clubroot. This can also be seen by the lower root index for plants treated with 100 ppm ProCa as well as the higher shoot index (Fig. 3). Both parameters indicate that the overall tolerance of the plant to the pathogen is higher. At all time points investigated, the galls were smaller than in untreated plants and more uninfected root was attached to the main gall (Fig. 6). Treatment with ProCa also resulted in a larger root system in uninfected plants. The effect was only prominent at 100 ppm ProCa, whilst lower dosages (25 ppm) had no effect. Higher concentrations of ProCa were also tried, but because of the effect on the upper part of the plants (dwarfish growth, data not shown) the higher concentration was not used in further experiments.

Figure 2.

Shoot phenotype of Arabidopsis thaliana plants after treatment with H2O (upper part) or prohexadione-calcium (ProCa) (lower part). Plants were photographed after inoculation with Plasmodiophora brassicae (23 days post-inoculation) or 37 days post-germination (equivalent age for uninfected plants). The bar represents 1 cm for all panels.

The mutant tt6 mimics the treatment with ProCa because it lacks flavanone-3-hydroxylase and consequently quercetin and kaempferol derivatives (Fig. 1). In tt6 a slight tolerance represented by a lower DI was found (Table 1).

Inhibition of flavonol synthesis in Arabidopsis with prohexadione-calcium

ProCa is an inhibitor for dioxygenases such as flavanone 3-hydroxylase (F3H) and gibberellin A20 3β-hydroxlase (Fig. 1), or 1-aminocyclopropane-1-carboxylic acid oxidase. F3H catalyses the conversion of naringenin to dihydrokaempferol and the conversion of eriodictyol to dihydroquercetin (Fig. 1). The inhibition of F3H by ProCa can result in the accumulation of naringenin and/or eriodictyol. However, eriodictyol can be converted to luteoforol if an unspecific flavanone reductase is present (Fig. 1).

It was tested whether ProCa treatment of Arabidopsis resulted in alterations of patterns for the major flavonoids naringenin, quercetin and kaempferol, which could be one reason for changes in clubroot development (Fig. 3). The levels of these flavonoids were determined by HPLC. In addition, the glycosylated compounds were determined after acid hydrolysis (Fig. 4). After root treatment with 100 ppm ProCa, kaempferol and quercetin were hardly detectable in any of the analysed samples, as expected if ProCa inhibits F3H. In contrast, naringenin accumulated to much higher levels (c. 5-fold) compared with non-treated roots. An increase in total naringenin after ProCa treatment was present in control and P. brassicae-infected roots, and was more due to the increase in the aglycone. The accumulation of naringenin also indicates that this compound is probably not further converted to other flavonoids, as observed in other plant species, or that such additionally synthesized flavonoids are present only in trace amounts.

Figure 3.

Detailed phytopathological analysis of infected roots with or without prohexadione-calcium (ProCa) treatment. A low root index (Ri/Rni) means small galls, a high shoot index (Si/Sni) means vigorous plants. The total root length (Lri/Lrni) was affected by low concentrations of ProCa.

Figure 4.

Accumulation of flavonoids in Arabidopsis thaliana control roots (data from Päsold et al., 2010; the data were recorded at the same time) and root galls after infection with Plasmodiophora brassicae and treatment with prohexadione-calcium (ProCa) determined by HPLC analysis. The control samples were taken 37 days post-germination, and the infected samples at 23 days post-inoculation (equivalent age to control plants). Upper panel: control roots and Plasmodiophora brassicae inoculated roots; lower panel: same tissues but plants treated simultaneously with 100 ppm ProCa.

One possible explanation of the effect of ProCa on the biosynthetic pathways, in addition to being a structural mimic of a cofactor, might be the influence on transcription by induction or repression of transcripts encoding biosynthetic enzymes. Using the expression database of the Arabidopsis eFP browser, indications were found that ProCa does not obviously influence any major transcript involved in the biosynthesis of flavonoids. In Figure 5, red coloured seedlings indicate up-regulated genes, blue colour down-regulated. Based on this finding it can be hypothesized that the effects described above and in the following sections are based on the enzymatic inhibition of the flavanone 3-hydroxylase by ProCa as an antagonist of the cofactor 2-oxoglutaric acid or alternative, as yet unknown, mechanisms.

Figure 5.

Transcript analysis for phenylpropanoid and flavonoid biosynthesis genes after treatment with 10 μm prohexadione-calcium (ProCa) taken from the Arabidopsis eFP browser (http://bar.utoronto.ca; Winter et al., 2007). The first plant shows the time point after 3 h treatment, the second after 12 h. Red indicates up-regulation of genes by ProCa, blue down-regulation. Abbreviations: PAL, phenylalanine ammonium-lyase; C4H, cinnamic acid 4-hydroxylase; 4CL, 4-coumarate-CoA ligase; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone-3-hydroxylase; F3′H, flavonoid-3′-hydroxylase; F3′5′H, flavonoid-3′5′-hydroxylase; (F)LS, flavonol synthase; DFR, dihydroflavonol reductase; LDOX, anthocyanin synthase; UF3GT, flavonol-3-O-glucosyltransferase (F3OG).

Effect of prohexadione-calcium on auxin responsiveness

Because the accumulation or absence of flavonoids is linked to auxin transport and, as a consequence, auxin response and accumulation (Buer & Muday, 2004), it was tested whether a different pattern of an auxin-responsive promoter in roots can be observed. Control roots and P. brassicae-inoculated plants harbouring the auxin-responsive DR5::GUS construct were treated with either H2O or ProCa, and these four combinations were investigated for altered expression patterns. First, it was shown that the root systems are more vigorous, as already indicated in the phytopathological analysis (Fig. 3; Table 1). Secondly, using the DR5::GUS line it was shown that auxin responsiveness was not changed in clubroots after treatment with ProCa (Fig. 6). The GUS activity was confined to the root parts harbouring the galls, independent of the treatment (Fig. 6), even though the root system that had not yet transformed into galls after ProCa treatment did not show strong GUS activity.

Figure 6.

Pictures of typical root galls and control roots with and without prohexadione-calcium (ProCa) treatment at three time points after inoculation (14, 18 and 21 days post-inoculation, dpi, for infected plants and 28, 32 and 35 days post-germination, dpg, for uninfected plants). The distribution of DR5::GUS staining is shown which indicates responsiveness to IAA. The bar represents 5 mm in all panels.

Effect of inhibition of gibberellin biosynthesis on clubroot formation

The group of dioxygenases inhibited by ProCa involves the flavonoid, but also gibberellin and ethylene biosynthesis. Therefore, a gibberellin biosynthesis inhibitor with a different target (Fig. 1) was also tested, which should not influence flavonoid synthesis (Table 1). The compound chlormequatchloride (CCC) was administered in the same manner as ProCa and the disease index (DI) determined. The plants treated with 100 ppm CCC did not show any gall size reduction, indicating that GA biosynthesis is not necessary for clubroot development. This also shows that ProCa does not regulate gall size by decrease of GA biosynthesis.

Discussion

Clubroot disease in the Brassicaceae alters many biosynthetic pathways (Siemens et al., 2006; Ludwig-Müller et al., 2009). It was previously shown that flavonoids accumulate in root galls and play a role in gall formation (Päsold et al., 2010). The most probable explanation given was the modulation of auxin efflux, thereby increasing local auxin levels in roots, which would in consequence lead to larger galls. This hypothesis was corroborated by investigating flavonoid biosynthesis mutants, of which some showed altered clubroot formation (Päsold et al., 2010).

Flavonoids generally play a significant role as defence compounds. For example, it was shown that flavonoid-type phytoalexins inhibit fungal growth (Dixon, 2001). The growth regulator ProCa was shown to influence the flavonoid biosynthesis pathway in different plant species by inhibition of the enzyme flavanone-3-hydroxylase (Puhl et al., 2008), so that phytoalexin-like substances accumulated, which led to more pathogen and pest tolerant plants (Rademacher et al., 2006). The aim of this study was to find out whether ProCa treatment could result in a reduction of the clubroot disease of Arabidopsis, because of altered flavonoid patterns. After ProCa application, clubroot symptoms were indeed reduced, indicated by a lower disease index, larger total root system, and an increase in the shoot index (Fig. 3; Table 1). To the authors' knowledge, this is the first report on the effect of ProCa on a root disease, as previous work concentrated on leaf pathogens (Halbwirth et al., 2003; Spinelli et al., 2005). ProCa has not previously been used on root diseases in practical applications, mainly due to its instability in soil (W. Rademacher, BASF, Limbugerhof, Germany, personal communication).

In the mutant tt6 the F3H is genetically altered, so that these plants do not accumulate wildtype amounts of quercetin and kaempferol and therefore should be a genetic mimic of ProCa treatment. Even though the disease index is reduced in the tt6 mutants (Table 1), the observed effects of ProCa on Arabidopsis roots cannot be explained by this mutant phenotype alone. For example, after ProCa treatment even the infected roots were larger (Fig. 6), whereas this was not the case for tt6 plants (data not shown). This might be due to the observation that the mutation in F3H is leaky (Peer et al., 2001). For the formation of high levels of the antimicrobial flavonoids the complete inhibition of F3H seems necessary. To mimic the effect of ProCa on apple leaves, transgenic antisense F3H apple plants were produced (Flachowsky et al., 2011). Both ProCa and silencing of F3H led to the accumulation of flavanones in apple leaves, but not to the formation of antimicrobial 3-deoxyflavonoids. Accordingly, no resistance to fire blight was observed for the antisense plants.

In order to resolve a possible mechanism for the reduction in clubroot disease, several experiments were carried out. First, the flavonoid levels of the three major compounds in Arabidopsis were determined. In apple (Halbwirth et al., 2003; Spinelli et al., 2005) and grapevine (Puhl et al., 2008), treatment with ProCa resulted in strongly changed flavonoid patterns, for example the accumulation of luteoliflavane. However, in Arabidopsis, naringenin accumulated strongly (Fig. 4), suggesting that the pathway is not altered to produce other compounds, or that these flavonoids are present only in trace amounts. These data suggest that Arabidopsis does not possess an unspecific enzyme for the conversion of flavonones to phytoalexin-type flavonols as was found in other plant species (Bazzi et al., 2003; Spinelli et al., 2005; Rademacher et al., 2006).

Secondly, because flavonoids are involved in the modulation of polar auxin transport (Buer & Muday, 2004), it was investigated whether ProCa treatment altered the auxin response using the promoter-reporter line DR5::GUS. If polar auxin transport were influenced by flavonoids and ProCa, changes in auxin distribution would be expected, which could be measured indirectly using the auxin-responsive promoter-reporter line. It was shown that auxin responsiveness was confined to the root parts harbouring the galls in all treatments, so that an effect of ProCa directly or via altered flavonoid patterns on auxin can be ruled out (Fig. 6).

Thirdly, the specificity of ProCa on flavonoid synthesis was tested by using another growth regulator, chlormequatchloride (CCC), inhibiting specifically gibberellin biosynthesis (Rademacher et al., 2006). In general, plant growth promoting hormones are suggested to be important for clubroot disease (Ludwig-Müller et al., 2009). The plants treated with CCC did not show any gall size reduction, showing for the first time experimentally that intact GA biosynthesis is not necessary for clubroot development and that low GA doses, which were shown to promote root growth (Tanimoto, 2012), do not account for the effect on clubroot reduction. ProCa also targets the last step in ethylene synthesis, aminocyclopropanecarboxylic acid oxidase (Rademacher et al., 2006). For the ethylene pathway it was recently shown that elevated ethylene levels did not alter disease development in Arabidopsis, while ethylene perception was necessary to restrict gall growth (Knaust & Ludwig-Müller, 2013), making it unlikely that ProCa has an effect on the ethylene biosynthetic pathway in Arabidopsis.

Are there other possible mechanisms of action for ProCa? Perhaps the calcium ions could be effectors? It cannot be determined whether Ca2+ is released from ProCa. However, in the complex management of clubroot in the field, calcium treatment together with acidic pH values plays a major role (Donald & Porter, 2009). The short period between germination of resting spores and penetration of zoospores is the phase where survival could be affected by soil factors, among them calcium. These factors influence the inoculum potential and its viability and invasive capacity (Dixon, 2009). Dixon & Webster (1988) demonstrated that treatment with calcium resulted in the inhibition of the total numbers of root-hair infections and the rate of maturation through plasmodial, sporangial and zoosporangial stages compared to controls. Elevated concentrations of calcium completely inhibit the later stages of P. brassicae development in the root hair (Dixon & Webster, 1988; Webster & Dixon, 1991; Donald & Porter, 2004). The work from Niwa et al. (2008) provided direct evidence that spore germination and subsequent root-hair colonization was retarded by the presence of calcium and alkaline pH values. Because calcium is a signal affecting many plant responses and is also involved directly in disease resistance (Romeis et al., 2001; Navazio et al., 2007), stimulation of the plant's defence responses by calcium (or the complex with prohexadione) cannot be ruled out.

In conclusion, it was shown that ProCa is an inhibitor of clubroot formation. However, a direct inhibitory effect of naringenin can be ruled out, because treatment of clubroots with naringenin did not result in the reduction of disease (Päsold et al., 2010). Either the two observations are not linked, or endogenous changes in naringenin have a different effect than application of the compound. Perhaps the ProCa-treated root systems are larger and therefore might simply be more robust against pathogens. A more general role of ProCa on roots can be further tested by using other root pathogens of Arabidopsis.

Acknowledgements

The authors would like to thank Dr Wilhelm Rademacher, BASF, for the gift of growth regulators and helpful discussions. The technical assistance of Sabine Rößler is also acknowledged.

Ancillary