Manipulation of plant innate immunity and gibberellin as factor of compatibility in the mutualistic association of barley roots with Piriformospora indica


*(fax +49 641 9937499; e-mail


Fungi of the order Sebacinales (Basidiomycota) are involved in a wide spectrum of mutualistic symbioses with various plants, thereby exhibiting unique potential for biocontrol strategies. Piriformospora indica, a model organism of this fungal order, is able to increase the biomass and grain yield of crop plants, and induces local and systemic resistance to fungal diseases and tolerance to abiotic stress. To elucidate the molecular basis for root colonization, we characterized the interaction of P. indica with barley roots by combining global gene expression profiling, metabolic profiling, and genetic studies. At the metabolic level, we show that fungal colonization reduces the availability of free sugars and amino acids to the root tip. At the transcriptional level, consecutive interaction stages covering pre-penetration-associated events and progressing through to root colonization showed differential regulation of signal perception and transduction components, secondary metabolism, and genes associated with membrane transport. Moreover, we observed stage-specific up-regulation of genes involved in phytohormone metabolism, mainly encompassing gibberellin, auxin and abscisic acid, but salicylic acid-associated gene expression was suppressed. The changes in hormone homoeostasis were accompanied with a general suppression of the plant innate immune system. Further genetic studies showed reduced fungal colonization in mutants that are impaired in gibberellin synthesis as well as perception, and implicate gibberellin as a modulator of the root’s basal defence. Our data further reveal the complexity of compatibility mechanisms in host–microbe interactions, and identify gibberellin signaling as potential target for successful fungi.


Despite the intensive measures taken to protect crops from diseases and pests, recent evaluations have shown that continuously increasing total crop production is accompanied by increased yield losses due to biotic and abiotic stresses (Oerke and Dehne, 2004; Lobell and Field, 2007). One solution to this problem is to improve crop production strategies to make them more reliable for the producer and safer for consumers and the environment (Cook, 2006). A key to this is to increase knowledge of the intricate and dynamic communications between crop plants and their interacting parasitic or beneficial microbial partners (Khush, 2005). By elucidating ‘compatibility mechanisms’, i.e. mechanisms of either disease or beneficial symbiosis development and the molecular networks supporting microbial virulence, key processes can be identified and exploited to develop more sustainable measures based on either novel chemicals or genetically improved crop plants.

In nature, plants are generally colonized by a range of fungal microbes that may have detrimental, neutral or beneficial effects on their hosts. For a unifying and balanced view on compatibility mechanisms, it is essential to study the parasitic lifestyles of biotrophs and hemi-biotrophs as well as those of mutualistic fungi (Kogel et al., 2006; Kogel, 2008; and references therein). Compatibility in host–microbe systems depends on biochemical interplay between molecules of the interacting partners, resulting in host recognition, host invasion, microbial nutrition, host colonization and microbial reproduction (Vögele and Mendgen, 2003; Hückelhoven, 2005, 2007; O’Connell and Panstruga, 2006; Robert-Seilaniantz et al., 2007; Speth et al., 2007). More specifically, conserved microbe-associated molecular patterns (MAMPs) and microbe-induced molecular patterns (MIMPs), lead to recognition of invaders by the plant (Jones and Dangl, 2006; Bent and Mackey, 2007). MAMP/MIMP recognition is achieved by plasma membrane-localized pattern recognition receptors (PRRs) initiating MAMP-triggered immunity (MTI). Successful in planta development of biotrophic and hemi-biotrophic pathogens and most probably microbial symbionts is entirely dependent on the release of effector molecules that specifically interfere with MTI and result in the phenomenon called effector-triggered susceptibility (Jones and Dangl, 2006). This distinct early phase of plant defence suppression is followed by a second phase of effector-mediated metabolic reprogramming of the host tissue that eventually results in successful microbial establishment (Cui et al., 2005; Göhre and Robatzek, 2008).

The root-colonizing basidiomycete Piriformospora indica is the archetype of the recently established mycorrhizal order Sebacinales (Weiss et al., 2004). Hallmarks of the mutualistic symbioses formed by these fungi with a broad range of mono- and dicotyledonous plants are growth promotion, yield increases, enhanced resistance to root and leaf pathogens, and abiotic stress tolerance (Waller et al., 2005; Deshmukh and Kogel, 2007; Shahollari et al., 2007; Stein et al., 2008). The colonization patterns of the various root regions show some qualitative differences that distinguish P. indica from obligate biotrophic arbuscular mycorrhizal fungi. The highest fungal biomass was found in the differentiation and root hair zones, and the meristematic zone was less extensively colonized (Deshmukh et al., 2006). In contrast, arbuscular mycorrhizal fungi are known to preferentially colonize younger root parts, as physiological activity of host cells is a prerequisite for efficient nutrient exchange between the symbiotic partners (Karandashov and Bucher, 2005). Indeed, one of the main qualitative differences between arbuscular mycorrhizal fungi (Glomeromycota) and P. indica mycorrhiza (Basidiomycota) is the dependence on cell death for root colonization at late interaction stages (>5 days after inoculation). However, this cell death-associated colonization does not lead to root necrotization as seen for hemi-biotrophic or necrotrophic fungi. Therefore, the term necrotrophy is misleading, and ‘cell death-dependent colonization’ is a more precise description of this interaction phase (Schäfer and Kogel, 2009). The dependence on host cell death was also shown in barley plants constitutively over-expressing the negative cell death regulator BAX INHIBITOR-1. As a result of the genetically increased cell viability, fungal root colonization was significantly reduced in these transgenic plants (Deshmukh et al., 2006). However, recent transmission electron microscopic analyses have revealed an initial biotrophic phase preceding the cell death-dependent colonization stage (P. Schäfer and B. Zechmann, unpublished results).

In the present study, we have assessed the response of barley roots to P. indica colonization by transcriptional and metabolic profiling. The most significant changes were observed in genes associated with signal perception and transduction, secondary metabolism, plant defence and hormone metabolism. These studies revealed complex interplay of P. indica with its host, during which gibberellin may be recruited to manipulate plant defence and to initiate the mutualistic symbiosis.


The transcriptome reflects a biphasic colonization of barley roots by Piriformospora indica

In an initial microscopic study, extracellular colonization of roots was seen within 1–2 days after inoculation (dai) with P. indica chlamydospores, during which fungal hyphae frequently fused in order to form an initial extracellular network. By 3 dai, intercellular hyphae were visible and single rhizodermal cells were penetrated without the formation of specific penetration organs. By 7 dai, large areas of the root surface were covered with P. indica mycelium, and inter- and intra-cellular hyphae were abundant in the rhizodermis and cortex. Fungal sporulation was most frequently initiated at approximately 14 dai (Figure 1). Root colonization was generally not accompanied by the emergence of structural and biochemical defence barriers, and did not coincide with tissue necrotization even at later cell death-associated interaction stages.

Figure 1.

 Schematic overview of barley root colonization by Piriformospora indica.
After chlamydospore germination (at approximately 12 h after inoculation), the fungus started to penetrate rhizodermal cells and intercellularly colonize the root cortex (3 dai). Subsequently, the fungus infests the root extra-, inter- and intra-cellularly. At approximately 7 dai, the fungus builds inter-/intra- and extra-cellular networks. Fungal sporulation is most frequently observed at approximately 14 dai. Fungal structures were stained using WGA-AF 488.

The time points 1, 3 and 7 dai were chosen for further analyses, as distinct interaction stages were covered: extracellular fungal development (1 dai), penetration-associated and early colonization events (3 dai), and progressive root cell colonization (7 dai). For transcriptome profiling, a custom-made 44K Agilent microarray was designed, consisting of approximately 40 000 probe sets (see Experimental procedures). Of these, 392 (1 dai), 459 (3 dai) and 509 (7 dai), respectively, were differentially regulated P. indica-colonized roots compared to mock-treated roots (Figure 2), while 1107 genes were differentially regulated at one of the three time points at least [Table S1, complete data accessible at the National Center for Biotechnology Information Gene Expression Omnibus (GEO) database (Edgar et al., 2002), accession GSE13756]. However, fewer than 10% of the identified genes at 1 dai showed altered expression at 3 (9%) and 7 dai (9.5%). In contrast, more genes displayed an overlapping expression pattern between 3 and 7 dai (approximately 40–50%). Interestingly, although approximately 50% of the genes were induced or suppressed at 1 dai, approximately 75% of genes were induced and only 25% suppressed at 3 and 7 dai.

Figure 2.

 Number of Piriformospora indica-responsive barley root genes.
Numbers of differentially regulated genes after root inoculation with fungal chlamydospores at 1, 3 and 7 dai, displayed as a Venn diagram. Light grey, induced genes; grey, suppressed genes; black, genes that are induced at one time point but suppressed at the other given time point.

The Agilent array data were verified by quantitative PCR with gene-specific primers for the genes encoding terpene synthase 7, syn-copalyl diphosphate synthase and a putative abscisic acid-induced protein (genes that showed broad variations in induction/suppression levels; Figure S1).

P. indica-colonized roots displayed pronounced alterations in expression of genes involved in stress responses

Annotation of differentially regulated genes resulted in 15 functional groups and two groups comprising either unknown ESTs (380 genes, 34.3%) or genes that could not be assigned to any group (34 genes, 3.1%) (Figure 3, Table 1 and Table S1).

Figure 3.

 Functional categories of genes in barley roots that were differentially regulated upon Piriformospora indica colonization.
Coloured bars illustrate the absolute number of genes that were down- or up-regulated within the various categories at 1, 3 and 7 dai. See inset for colour code.

Table 1.   Distribution of differentially regulated genes within each functional category
 Number of genesPercentage of genes1 dai3 dai7 dai1 dai3 dai7 dai
UpDownUpDownUpDownPercentage of genesaPercentage of genesaPercentage of genesa
  1. aTotal % of genes regulated at indicated time points.

Cell wall metabolism211.95128421.52.21.2
Cellular traffic/cytoskeleton201.83132751.01.12.4
Defence/stress response15113.62518698691111.016.815.7
DNA metabolism/genome organization232.19433423.31.31.2
Hormone metabolism383.4951851333.65.03.1
Lipid metabolism363.3551341632.63.73.7
Nutrient storage131.27131012.00.90.2
Primary metabolism333.048102843.12.62.4
Protein degradation343.1581421333.33.53.1
Protein metabolism333.096621223.81.72.8
Secondary metabolism585.21183452744.88.56.1
Transcription/protein biosynthesis484.311111351165.63.93.3
Total percentage57.442.676.923.173.726.3 

Genes involved in plant defence/stress responses represent the largest group of differentially regulated genes (151 genes, 13.6%). Likewise, many signalling components were affected by P. indica (90 genes, 8.1%). Genes participating in secondary metabolism (58, 5.2%), those encoding transporters/channels/pumps (53, 4.8%), and those involved in transcription/protein biosynthesis (48, 4.3%) also showed pronounced transcriptional alteration. In contrast, nutrient storage (13, 1.2%), cellular trafficking/cytoskeleton (20, 1.8%), cell-wall metabolism (21, 1.9%) and DNA metabolism/genome organization (23, 2.1%) were only weakly influenced.

Most functional groups showed a stage-dependent expression profile. Genes associated with transcription/protein biosynthesis and signalling were strongly transcriptionally altered at 1 dai. Components of secondary metabolism and transporters/channels/pumps showed the greatest differences at 3 dai. Finally, transcripts of receptors and proteins involved in plant defence/stress responses exhibited a higher degree of differential regulation at 3 and 7 dai (Table 1).

P. indica interferes with plant defence and affects signal perception and transduction

The diverse set of defence/stress-responsive genes induced or suppressed by P. indica encoded ‘defence-related’ proteins (e.g. R proteins, PR proteins) as well as genes encoding ‘stress-responsive’ enzymes (e.g. laccases, late-embryogenesis-abundant proteins/dehydrins), indicating that P. indica elicited a rather non-specific defence reaction. As approximately 4% of all ESTs on the array were defined as defence/stress-related, their pronounced differential regulation cannot be explained by their exceeding presence on the array. At 1 dai, 11% of differentially regulated genes (25 induced/18 suppressed) were classified in this category, while 16.8% (69 induced/eight suppressed) and 15.7% (69 induced/11 suppressed) were identified at 3 and 7 dai, respectively. In total, only five genes encoding two putative PR10s, a putative laccase 18, a putative germin A and a putative syringolide-induced protein were differentially regulated at all time points. Based on their expression pattern, the 151 defence/stress-responsive genes were divided into four regulation clusters (Figure S2a–d). Cluster A consists of genes that were suppressed by P. indica (35 genes, 23%). Genes in cluster B showed a transient induction profile at 1 or 3 dai (47 genes, 31%). All genes that showed transient induction at 3 dai but lower up-regulation at 7 dai (12 genes, 8%) were assigned to cluster C. Finally, cluster D encompasses genes that were steadily up-regulated (31 genes) at 3 and 7 dai or exclusively induced at 7 dai (26 genes). Based on their expression pattern, the genes of cluster D might exclusively code for proteins that effectively restrict root colonization, and several germins, which are known to restrict powdery mildew infection of barley leaves (Zimmermann et al., 2006), were represented in this cluster. Alternatively, cluster D might include genes involved in regulation of cell death, which is frequently observed at 7 dai (Deshmukh et al., 2006), or genes that are activated by cell death-derived signals released by dying cells. However, of these 57 genes, 35% showed a fold change induction >4. This is in accordance with the generally moderate induction level of defence/stress-associated genes: approximately 70% of the genes showed a less than fourfold induction level at all time points. Interestingly, the highest induction values were found at the pre-penetration stage (1 dai) (Figure S2).

A high number of differentially abundant transcripts encoded receptors (42 ESTs) and signal transducers (90 ESTs) (encompassing transcription factors, DNA-binding proteins and protein kinases) (Table S1). Again, receptor gene expression overlap was mainly observed between 3 and 7 dai (52–68%), and to a minor extent at 1 dai (10–20%). Similarly, approximately 45% of the genes involved in cell signalling showed congruent expression between 3 and 7 dai.

P. indica-induced changes in auxin, ABA, and brassinosteroid synthesis and signalling

Genes encoding for a tryptophan decarboxylase and a putative indole-3-glycerol phosphate synthase involved in l-tryptophan synthesis, an immediate precursor of IAA (Ljung et al., 2005), were up-regulated at 3 and 7 dai. In addition, a second tryptophan decarboxylase and a putative anthranilate phosphoribosyl transferase that might be involved in tryptophan synthesis show maximum expression at 3 dai (Figure 4). Further, an auxin-induced protein and a flavin-containing mono-oxygenase (YUCCA3) were up-regulated and an auxin-repressed protein was down-regulated at 3 dai and/or 7 dai suggesting that auxin biosynthesis and signalling might be activated during symbiotic colonization.

Figure 4.

 Changes of transcripts involved in hormone signalling during Piriformospora indica colonization.
Genes involved in hormone metabolism and responses that are differentially regulated at one time point at least are shown. Colours represent fold changes of each gene, which are either up-regulated (red) or down-regulated (blue) compared to mock-inoculated roots. Fold changes (FC) were calculated by dividing antilog signal intensities obtained from arrays hybridized with cDNA of mock- and P. indica-ztreated roots.

Changes in P. indica-induced hormone balance were also seen for the sesquiterpenoid ABA (Figure 4). The hormone plays a crucial role in abiotic and biotic stress responses (Finkelstein and Rock, 2002), and often shows antagonistic activity for other hormones (Asselbergh et al., 2008). Four ABA-responsive proteins of unknown function were induced at 1 dai, but repressed at later time points. The identification of several genes encoding late embryogenesis abundant (LEA) proteins/dehydrins might also be due to ABA accumulation, as several members of that gene family are ABA-responsive (Hundertmark and Hincha, 2008).

The gene encoding cycloartenol synthase (CS), which that contributes to the synthesis of brassinosteroid (BR) precursors, and BLE2, which encodes a BR-responsive nine transmembrane protein, were induced at 1 dai. In addition, two BAK1 genes encoding brassinosteroid insensitive 1-associated receptor kinases 1 that are involved in BR signalling were induced at 3 and 7 dai.

P. indica modifies the expression of genes involved in oxylipin synthesis

Root colonization by P. indica is associated with transcriptional changes in genes associated with lipid metabolism (Table S1). Hydrolysis of phospholipids by lipases leads to the release of unsaturated fatty acids, which can serve as substrates for the synthesis of oxylipins (Feussner and Wasternack, 2002; Meijer and Munnik, 2003; Shah, 2005). The micoarray analysis indicated that genes encoding four oleate Δ12-desaturases that convert oleate to linoleic acid were differentially regulated (three probe sets at 3 and 7 dai, and one transcript at 3 dai) and a gene encoding cytochrome b5, which is required as an electron donor for desaturation, was similarly induced (see lipid metabolism, Table S1). Central to oxylipin synthesis is the action of lipoxygenases (LOXs) that convert linoleic or α-linolenic acid to the oxylipin precursors (9S)- and (13S)-hydroperoxide (Feussner and Wasternack, 2002). Two LOX genes, LOX2.1 and LOX2, which catalyse the oxidation of linoleic acid to 13-hydroxyoctadecadienoic acid (Peng et al., 1994; Vörös et al., 1998), were induced at 3 and 7 dai (Figure 4). Interestingly, oxylipin synthesis is apparently suppressed at the pre-penetration phase (1 dai), as LOX2.2 and a gene encoding a jasmonate-induced protein (involved in the downstream JA response) were found to be suppressed at 1 dai. Suppression of JIP23 encoding 23 kDa jasmonate-induced protein at 3 dai might suggest synthesis of oxylipins other than jasmonate.

Alterations in the methylerythritol phosphate (MEP) pathway and synthesis of secondary metabolites

Terpenoids derive from the C5 precursors isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP), which are synthesized either via mevalonate or the MEP pathway. Almost all genes of the MEP pathway are induced by P. indica at late stages (Figure 5). Genes encoding two putative geranylgeranyl diphosphate synthases, which produce geranylgeranyl diphosphate (GGDP) from IPP and DMAPP, were induced at 3 and 7 dai. GGDP, in turn, is the precursor for mono-, di- and sesquiterpenes, the family to which anti-microbial phytoalexins, carotenoids or phytohormones such as gibberellins (GA) or ABA belong. Downstream of GGDP, the most strongly induced gene was one encoding a terpene synthase (44-fold at 3 dai; see also Figure S1). In parallel, a gene encoding a putative syn-copalyl diphosphate synthase (syn-CDS) mediating the cyclization of GGDP and a 10-deacetylbaccatin III-10-O-acetyl transferase-like gene, whose homologue is associated with taxol synthesis in Taxus x media (Guo et al., 2007), were induced at all time points. A high number of cytochrome P450 mono-oxygenases of unknown function were also induced by P. indica. Various members of this enzyme family are involved in the production of both diterpene phytoalexins (Okada et al., 2007) and GA (Yamaguchi, 2008) in rice.

Figure 5.

Piriformospora indica induces the methylerythritol phosphate (MEP) pathway and gibberellic acid (GA) synthetic genes in barley.
(a) Scheme of the MEP pathway and GA biosynthesis. Products/substrates: glyceraldehyde-3-phosphate (G3P); 1-deoxy-d-xylulose 5-phosphate (DXP); 2-C-methyl-d-erythritol 4-phosphate (MEP); 4-(cytidine 5′-diphospho)-2-C-methyl-d-erythritol(CDP-ME); 2-phospho-4-(cytidine 5′-diphospho)-2-C-methyl-d-erythritol (CDP-ME2P); 2-C-methyl-d-erythritol 2,4-cyclodiphosphate (MECDP); 1-hydroxy-2-methyl-2-(E)-butenyl 4-diphosphate (HMBDP); isopentenyl diphosphate (IPP); dimethylallyl diphosphate (DMAPP); geranylgeranyl diphosphate (GGDP); copalyl diphosphate (CDP); gibberellin A12 (GA12); gibberellin A3 (GA3); gibberellin A4 (GA4); gibberellin A8 (GA8); gibberellin A34 (GA34). Enzymes: DXP synthase (DXS); DXP reductoisomerase (DXR); CDP-ME synthase (CMS); CDP-ME kinase (CMK); MECDP synthase (MCS); HMBDP synthase (HDS); HMBDP reductase (HDR); IPP isomerase (IPI); GGDP synthase (GGPS); ent-CDP synthase (ent-CPS); ent-kaurene synthase (ent-KS); ent-kaurene oxidase (ent-KO); ent-kaurenoic acid oxidase (ent-KAO); gibberellin 3 oxidase (GA3ox); gibberellin 20 oxidase (GA20ox); gibberellin 2 oxidase (GA2ox).
(b) Genes involved in the MEP pathway and GA biosynthesis that are differentially regulated in P. indica-colonized barley roots at 1, 3 and 7 dai. Numbers indicate fold induction of respective genes in P. indica-colonized versus mock-treated roots.

As carotenoids represent one major product class of the MEP pathway, we examined whether transcriptional induction of this pathway and of geranylgeranyl diphosphate synthase (GGPS) would result in elevated carotenoid production in P. indica-colonized roots. We found that barley roots contained minute amounts of carotenoids. Of the 10 detected carotenoids, six low-abundance carotenoid species could be reliably quantified. Violaxanthin and neoxanthin were the most abundant of these, and are also precursors in ABA synthesis (Finkelstein and Rock, 2002). The violaxanthin content was slightly lower in inoculated compared to non-inoculated roots at 1 dai, but not at later stages (Figure 6i). The amount of total neoxanthin (cis- and trans-neoxanthin, Figure 6j), the major carotenoid in barley roots, was reduced at 1 dai but elevated at 7 dai, indicating increased productivity of the MEP pathway in proximal root segments.

Figure 6.

 Carbohydrate, amino acid and carotenoid contents in barley roots colonized by Piriformospora indica.
For analysis, the whole root (1 dai) or proximal 3 cm of the roots (3 and 7 dai) from P. indica-inoculated and control plants were analysed. Contents of hexoses (a), sucrose (b) and starch (c), total amino acid content (d), and contents of glutamine (e), asparagine (f), glutamate (g), aspartate (h), violaxanthine (i) and neoxanthine (j) in the proximal 3 cm of control roots (grey) and P. indica-colonized roots (black) at 1, 3 and 7 dai are shown. Data are the means of 12 independent samples from three independent experiments; error bars represent the standard error.

Barley mutants impaired in gibberellin synthesis or perception are less extensively colonized by P. indica

As GA is produced from GGDP, we searched for genes involved in GA biosynthesis downstream of GGDP, and found two differentially expressed genes encoding putative ent-kaurene synthases at 3 and 7 dai (Figure 5). Accordingly, a GA2ox gene mediating inactivation of active GA (Yamaguchi, 2008) was down-regulated in response to P. indica at 3 dai. These results suggest that GA biosynthesis is raised by P. indica.

In order to determine the impact of GA synthesis and signalling on P. indica colonization, we analysed barley GA mutants. M117 has a low endogenous GA content, probably caused by a block at either geranylgeranyl diphosphate synthase or syn-copalyl diphosphate synthase, as it fails to accumulate ent-kaurene in the presence of tetcyclacis (which prevents further oxidation to kaurenoic acid, J.R. Lenton and P.M.C., unpublished results; Chandler and Robertson, 1999). M121 is GA-insensitive due to a mutation in the gene encoding the GA receptor GID1 (Chandler et al., 2008). Significantly, both mutants showed reduced colonization by P. indica (Figure 7a), which was cytologically detectable as a reduced amount of fungal hyphae at 7 dai and reduced intracellular sporulation at 21 dai. However, structural defence responses (e.g. cell-wall fortifications) were not detected in either mutant. As GA has been shown to affect the balance between other phytohormones (Navarro et al., 2008), we monitored defence responses known to be associated with SA, JA and ethylene during P. indica colonization. Alterations in barley root GA homoeostasis were associated with elevated expression of PR10 and the SA-responsive gene PR1B at 3 and 7 dai in both mutants (Figure 7b). Similarly, the defence-associated gene PR5 was induced by the fungus at 3 dai. Ethylene-responsive RAF1 expression was not affected (data not shown).

Figure 7.

 Amount of fungal colonization of GA mutants and alterations in expression of defence genes in roots.
(a)Relative amount of fungal DNA in GA mutants at 3 and 7 dai as determined by quantitative PCR.
(b) Relative expression of PR1B, PR5, PR10 and JIP23 in roots of M117 and M121 in response to Piriformospora indica at 3 and 7 dai. The data are based on three independent biological experiments. The barley (cv. Himalaya) mutant lines used were impaired in GA synthesis (M117) or defective in the GA receptor GID1 (M121). Data were analyzed by analysis of variance (anova) using a block design. Asterisks indicate significance at < 0.05.

The availability of C and N assimilates is decreased in barley roots colonized with P. indica

In contrast to defence/stress-associated genes, phytohormone signalling and secondary metabolism, genes involved in primary metabolism showed only minor differences in transcript levels at all time points (Table 1). Nevertheless, we expected a shift in assimilate availability in response to P. indica colonization due to additional sink activity by the fungus. Therefore, we assessed the amounts of free sugars and amino acids. In general, differences in hexose, sucrose and amino acid contents followed a developmental pattern in the proximal parts of harvested roots (Figure 6a,b,d). Hexose and amino acid contents decreased sharply with increasing root age, and starch contents exhibited a slight decrease at 7 dai, but sucrose contents increased strongly from 1 to 3 dai. Phloem transport of amino acids such as glutamine and asparagine decreased by 90% from 1 to 3 dai (Figure 6e,f), as did that of many minor amino acids (data not shown). In contrast, transport of another group of amino acids (e.g. glutamate and aspartate) did not decrease as strongly (Figure 6g,h).

As the results of these experiments suggested a developmental decline in sugar and amino acid contents with increasing degree of differentiation of the harvested root portion, we determined the sugar and amino acid contents in distal root regions encompassing the calyptra and the meristematic (0–0.5 cm), elongation (0.5–1 cm) and differentiation zones (1–1.5 and 1.5–3 cm). Piriformospora indica colonization led to a decrease of hexose content in all sampled segments (Figure 8a), but no changes were observed for sucrose (Figure 8c). The sucrose/hexose ratio (Figure 8e) exhibited a stronger decrease in P. indica-colonized roots compared to mock-treated roots when moving from proximal root segments towards the root tip, indicating that the supply of sugars to the sink tissue at the root tip is diminished in the presence of P. indica. The starch content along the root axis followed a similar pattern to the hexose content (Figure 8g).

Figure 8.

 Carbohydrate and amino acid contents in segments of barley roots colonized by Piriformospora indica.
For analyses, roots were divided into four segments (0–0.5 cm, calyptra and meristematic zone; 0.5–1 cm, elongation zone; 1.0–1.5 and 1.5–3 cm, differentiation zones), and were harvested from P. indica-colonized plants and control plants at 3 dai. The contents of hexoses (a) and sucrose (c), the sucrose/hexose ratio (e) and the starch content (g), as well as the total amino acid content (b) and contents of glutamine (d), asparagine (f), glutamate (h) and aspartate (i) in control roots (grey) and P. indica-colonized roots (black) are shown. The results shown are those of one representative experiment out of two. Data are the means of four independent samples, and error bars represent the standard error.

Similar to hexose and starch, P. indica-colonized roots contained fewer total free amino acids in the root tip and elongation zone (Figure 8b), due to a decrease in glutamine (Figure 8d), asparagine (Figure 8f) and glutamate (Figure 8h). In contrast, the higher total amino acid contents in the differentiation zone upon colonization with P. indica resulted from increased amounts of glutamate and aspartate (Figure 8h,i).


Regulation of defence/stress-related genes

Although stress-related genes represent the largest group among P. indica-induced genes in barley roots (151 genes), induction levels are generally moderate (Figure 3, Figure S2 and Table 1). In addition, only 31 genes of cluster D were constitutively induced, while the remaining genes were suppressed or displayed a transient induction (Figure S2). In general, the defence/stress-responsive genes affected by P. indica are associated with abiotic as well as biotic stress. This gene spectrum is rather broad and reminiscent of MTI responses that are non-specific and moderate in terms of the activation level of responses (Zipfel et al., 2004; Jones and Dangl, 2006; Wan et al., 2008). Over time, the gene spectrum alters most significantly between 1 dai and the later time points, while approximately 50% of the differentially regulated genes overlap between 3 and 7 dai. These expression profiles most probably reflect extracellular fungal development at 1 dai compared to inter-/intra-cellular colonization at the later time points. The high induction levels of defence/stress-responsive genes at 1 dai might indicate recognition of the fungus by the plant innate immune system. It is reasonable to speculate that fungal MAMPs (e.g. chitin) lead to defence activation. In turn, reduced induction or even suppression of respective genes at 3 and 7 dai might indicate active manipulation of the plant surveillance system and respective defence signalling cascades by the fungus. The differences recorded between 3 and 7 dai corroborate cytological studies that indicated an initial biotrophic phase followed by a cell death-dependent phase (P. Schäfer and B. Zechmann, unpublished results). Using transmission electron microscopy, P. indica was shown to colonize living Arabidopsis root cells by invaginating the plant plasma membrane. As the interaction proceeded, colonized cells died. However, adjacent non-colonized root cells were not affected or impaired in viability, suggesting that the fungus does not release cytotoxic molecules in order to kill cells ahead of penetration. As later colonization stages were not accompanied by tissue necrotization, it is more appropriate to use the term ‘cell death-dependent’, instead of the ‘necrotrophic colonization phase’ (Schäfer and Kogel, 2009). In conclusion, those genes induced or suppressed at 7 dai but not at 3 dai might participate in cell-death regulation or their expression might be modified by signals originating from dying cells. Liu et al. (2005) showed that cell death-derived signals might be translocated in neighbouring cells and exhibit pro-apoptotic activity. These authors found that a malfunctional autophagy pathway did not restrict tobacco mosaic virus-induced hypersensitive response cell death to the initial infection site and resulted in a spreading cell death phenotype. In our study, among the genes with highest transcript abundance at 3 and 7 dai are several members of the germin multi-gene family, some of which are developmentally regulated in the roots and leaves of seedlings under non-stress conditions and are thought to function in cell-wall metabolism (Zimmermann et al., 2006). In barley leaves, GerA, Ger4c and Ger4d are strongly induced after powdery mildew attack, and contribute to fungal growth arrest (Zimmermann et al., 2006). Hence, germin induction in roots may restrict P. indica invasion.

It is appealing to speculate that genes categorized as ‘defence/stress-responsive’ support the plant in balancing the mutualistic colonisation by P. indica. For instance, two LysM receptor-like kinases were found to be induced. The extracellular lysin motifs of plant LysM receptor-like kinases signify such proteins as receptors (Zhang et al., 2007). Recently, CERK1 was identified, which encodes a LysM receptor-like kinase that participates in chitin recognition and MTI (Miya et al., 2007; Wan et al., 2008). In contrast, NFR1 and NFR5 from Lotus japonicus were identified as crucial components for rhizobial nodulation by binding to Nod factors released by N2-fixing bacteria (Limpens et al., 2003; Radutoiu et al., 2007). It remains to be investigated whether either or both LysM receptor-like kinases identified in our study support or restrict establishment of the sebacinoid symbiosis.

In addition, an Arabidopsis homologue of the two BAK1 genes identified in our study was previously shown to be involved in basal defence triggered by flagellin (Chinchilla et al., 2007). Hence, BAK1 induction might be involved in MTI responses triggered by P. indica rather than in brassinolide signalling.

Reduced GA synthesis represses root compatibility

Our microarray analyses revealed comprehensive induction of the MEP pathway, which delivers precursors for GA synthesis (Figure 5). This is in line with the induction of two putative ent-kaurene synthases that may be involved in GA synthesis, and suppression of GA2ox1, the product of which inactivates GA (Figure 5). Subsequent genetic studies revealed reduced colonization by P. indica of two mutants, M117 and M121, that are impaired in GA synthesis or perception (Figure 7a). These phenotypes might be partially explained by an altered defence response, as the mutants showed elevated expression of PR1, PR5 and PR10 (Figure 7b). Recently, Navarro et al. (2008) demonstrated that Arabidopsis mutants blocked in GA signalling show enhanced resistance against necrotrophic Alternaria brassicicola. In contrast, quadruple DELLA mutants that show constitutive GA signalling exhibited increased susceptibility (Navarro et al., 2008). Interestingly, the JA/ethylene-responsive gene PDF1.2 showed delayed expression in DELLA in response to A. brassicicola or methyl jasmonate treatment (Navarro et al., 2008). These results suggest a direct connection between GA signalling and SA/JA responses. In analogy altered GA homoeostasis might explain induction of PR1B in the barley–P. indica interaction. However, JIP23 expression was affected only marginally (Figure 7b). Further biochemical and genetic studies are required to elucidate at which level (synthesis, perception or signal transduction) P. indica affects GA chomoeostasis in barley roots, and to what extent altered compatibility in GA mutants is a consequence of a modified defence response (e.g. suppression of SA-triggered responses).

Impact of P. indica on salicylic acid, jasmonate and ethylene signalling

The phytohormones salicylic acid, jasmonate and ethylene are components of the plant innate immune system, and have a considerable impact on pathogenic as well as mutualistic interaction partners (Glazebrook, 2005; Loake and Grant, 2007). Interestingly, genes encoding enzymes of the phenylpropanoid pathway, which is involved in the synthesis of SA, phytoalexin and lignin precursors, are weakly or transiently induced at 3 dai or down-regulated by P. indica (Table S1), which is consistent with the results of cytological studies, which rarely showed cell-wall lignification during P. indica colonization (Schäfer and Kogel, 2009; P.S., unpublished results).

Root colonization by P. indica is also accompanied by altered expression patterns of genes that are known to participate in oxylipin metabolism. Lipoxygenases (LOXs) catalyse the dioxygenation of linoleic or α-linolenic acid to (9S)- and (13S)-hydroperoxides, which are precursors of various oxylipins (Feussner and Wasternack, 2002). As oxylipins can act in plant defence as bioactive messengers (Blee, 2002; Feussner and Wasternack, 2002), as anti-microbial compounds (Weber et al., 1999), or as cell death-promoting agents (Rusterucci et al., 1999; Vollenweider et al., 2000), P. indica-responsive LOXs might affect root colonization. The results for JA synthesis/signalling are contradictory, as a putative S-adenosyl-l-methionine:jasmonic acid carboxyl methyltransferase mediating methyl jasmonate synthesis was induced at 3 and 7 dai, but JA marker genes (e.g. JIP23) were suppressed at 3 dai. As the enzymatic activity of the putative S-adenosyl-l-methionine:jasmonic acid carboxyl methyltransferase has not yet been demonstrated, synthesis of certain oxylipins other than JA or methyl jasmonate might be induced by P. indica.

The microarray data paint a similar picture regarding ethylene synthesis and signalling at later interaction stages. Two ESTs encoding 1-aminocyclopropane-1-carboxylic acid oxidase (ACC oxidase), which is involved in ethylene synthesis, were induced by P. indica at 3 and/or 7 dai. In contrast, genes encoding two transcription factors (ethylene-responsive factor, RAV2-like DNA binding protein) and an ethylene binding protein-like gene were down-regulated at 3 or 7 dai. This contradiction in ethylene synthesis and signalling might indicate induction of other yet to be identified ethylene signalling components. Alternatively, the ACC oxidases might be post-transcriptionally or post-translationally inactivated, thereby preventing ethylene synthesis.

ABA and auxin may act as negative regulators of root innate immunity

In addition to initiation after de novo ABA synthesis, ABA signalling is facilitated by interaction of phospatidic acid with the repressor ABI1 (Zhang et al., 2004), and active ABA can be rapidly recruited from the glucose-conjugated ABA pool (Lee et al., 2006). ABA regulates expression of members of the dehydrins/LEA protein family, various members of which were strongly induced at 1 dai. Dehydrins/LEAs function as chaperone-like proteins and maintain cellular functions under stress conditions (Hundertmark and Hincha, 2008). In Arabidopsis, ABA mediates susceptibility against Pseudomonas syringae pv. tomato. Type 3 effectors released by P. syringae pv. tomato cause elevation of ABA and JA in leaves, thereby abolishing callose deposition and MTI (De Torres-Zabala et al., 2007). Previous studies have shown ABA-mediated reduction of lignin and SA synthesis, and suppression of the phenylpropanoid pathway and various defence-related genes (Ward et al., 1989; Mohr and Cahill, 2007). ABA has also been shown to suppress basal and JA/ET-related defences, while ABA deficiency led to enhanced resistance against Fusarium oxysporum in Arabidopsis (Anderson et al., 2004). Our data are consistent with the hypothesis that ABA signalling might be used by P. indica to overcome initial host defence and to prepare for cell penetration and host colonization.

In addition, genes participating in auxin signalling and synthesis were induced by P. indica. As auxin mediates lateral root initiation and formation (Ljung et al., 2005), and P. indica enhances lateral root formation and primary root emergence (S. Jacobs and A. Molitor, unpublished results), the induction of auxin biosynthetic genes at 3 and 7 dai might support plant growth. Genetic studies in Arabidopsis have further demonstrated reduced bacterial growth in plant mutants that are repressed in auxin signalling (Navarro et al., 2006). Interestingly, P. indica was also reported to produce auxin (Sirrenberg et al., 2007). Taken together, these results suggest that P. indica might increase auxin signalling in order to (i) change the root morphology, thereby improving root accessibility, and/or (ii) impair plant defence.

P. indica influences primary metabolite distribution in barley roots

When we assessed the major carbohydrate and amino acid contents at various time points after root colonization by P. indica, most changes could be attributed to metabolite gradients along the root axis rather than fungal colonization. There were no consistent effects on the transcriptional regulation of primary metabolism, which contradicts our initial assumption that the presence of the fungus would affect metabolite redistribution. We observed high hexose and amino acid contents at 1 dai (Figure 6), confirming that the meristematic zone in the root tip represents a strong sink tissue (Herbers and Sonnewald, 1998). The high starch content in the most distal root segment is due to the presence of amyloplasts in the calyptra. Because our data suggest that most differences between P. indica-colonized and control roots might be obscured by metabolite gradients along the root axis, we assessed metabolite contents in various root segments at 3 dai. We found a decrease in hexose, glutamine and asparagine contents from tip to base (Figure 8), supporting the view that the hexose and amino acid contents depend on root differentiation. The tips of colonized roots showed a decrease in hexose, glutamine and asparagine contents, suggesting that sink strength is decreased by P. indica, and the sucrose/hexose ratio, an indicator of lower sink strength, was higher in colonized segments. There are two possible explanations for this. First, the availability of assimilates transported via the phloem might be lower at the root tip due to competition with P. indica, which predominantly resides in the differentiation zone (Deshmukh et al., 2006). Uptake of hexoses and amino acids from the root has been demonstrated for arbuscular mycorrhizal fungi (Pfeffer et al., 1999; Govindarajulu et al., 2005), and it is known that host cells and symbiont can compete for carbon when the supply from the phloem is limiting (Son and Smith, 1988). Second, cell death (Deshmukh et al., 2006) and auxin synthesis/signalling correlated with root colonization, which might explain the initiation of lateral roots in response to the fungus. Therefore, the supply to the primary root tip could be lower due to increased distribution to competing lateral root primordia.

Experimental procedures

Plant and fungal material

For all experiments, barley seeds (Hordeum vulgare L. cv. Golden Promise, cv. Himalaya and GA mutants M117, M121) (Chandler and Robertson, 1999; Chandler et al., 2008) were surface-sterilized, pre-germinated, inoculated with chlamydospores or mock-treated as described previously (Deshmukh et al., 2006). For the transcriptome and metabolome experiments, seedlings were grown in a 2:1 mixture of Seramis expanded clay (Mars, and Oil Dri (Damolin, under 16 h light (60 mmol m−2 sec−1 photon flux density) at 22/18°C (day/night) and 60% relative humidity. Three independent biological experiments were carried out. Barley roots were harvested at 1, 3 and 7 dai by carefully removing the seedlings from the substrate. Because of the higher colonization of older root parts, the upper 3 cm of the root (next to the kernel base) were collected at 3 and 7 dai, and aliquots were quick-frozen in liquid nitrogen. At 1 dai, the whole root was harvested. For each sample, 96 plants were harvested and divided into four subsets, which were used for metabolome analyses. For the transcriptome analyses, the roots of the four subsets were pooled and used for RNA isolation. Aliquots of homogenized frozen root material were used to quantify fungal biomass in inoculated roots by quantitative PCR, and for metabolite analyses, RNA or DNA isolation (see below). For the GA mutant analyses and metabolome studies of apical root segments, inoculated plants were grown on modified plant nutrient medium (0.5 mm KNO3, 2 mm MgSO4, 0.2 mm Ca(NO3)2, 0.43 mm NaCl, 0.14 mm K2HPO4, 2 ml/l Fe-EDTA [20 mm FeSO4, 20 mm Na2EDTA]) under the same growth chamber conditions as described above. GA mutant roots were harvested 3 and 7 dai. For determination of metabolites in apical root segments, roots of P. indica-inoculated and mock-treated cv. Golden Promise were removed from 1l glass jars at 3 dai and dissected into four segments (the apical first 0.5, 0.5–1, 1–1.5 and 1.5–3 cm from the tip). Pooled material from the individual segments was separately shock-frozen in liquid nitrogen and analysed for metabolite content (see below).

For all experiments, roots were cytologically analysed after tissue fixation in trichloroacetic acid solution and staining with WGA-AF488 (Molecular Probes, for fungal colonization and the absence of fungal contaminants in mock-treated roots by epifluorescence microscopy as described previously (Deshmukh et al., 2006).

Quantification of fungal colonization by quantitative PCR

Genomic DNA of wild-type and GA mutant plants was extracted from approximately 100 mg root material using a plant DNeasy kit (Qiagen, according to the manufacturer’s instructions. Aliquots (10 ng) of total DNA were used as the template for quantitative PCR analyses. Amplifications were performed in 20 μl SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich, with 350 nm oligonucleotides, using an Mx3000P thermal cycler with a standard amplification protocol (Stratagene, The inline image method (Livak and Schmittgen, 2001) was used to determine the degree of root colonization. Cycle threshold (Ct) values were obtained by subtracting the raw Ct values for the P. indica Tef gene (Bütehorn et al., 2000) from the raw Ct values for plant-specific ubiquitin (see Table S2 for the specific oligonucleotide primers used). Data were analysed using the ‘lm’ statistical procedure (linear model) in R using a block design. The marginal means were compared for all variables (a, b, c).

Design of barley oligonucleotide arrays

The 444 652 barley EST sequences publically available at in June 2007 were assembled into 28 001 consensus sequences, leaving 22 937 singletons. The assembly is available on the HarvEST website (assembly 35, For calculation of a 44K 60-mer oligonucleotide array using the Agilent eArray algorithm (, 13 265 singletons with a significant hit (E-value <10−10) in Arabidopsis or rice and the 28 001 consensus sequences were used, together with 2600 replicate probes and internal controls. The oligonucleotide sequences spotted on the 44K array are available at The latest annotation can be obtained at

Transcriptome analysis

For transcriptome studies, RNA was extracted from homogenized root material using TRIzol (Invitrogen, as described by the manufacturer. Aliquots (1 μg) of total RNA were used for cDNA synthesis with a qScript cDNA synthesis kit (Quanta Biosciences, RNA quality was analysed using an Aligent 2100 bioanalyser (Agilent, Probe synthesis and labelling were performed according to Agilent’s protocol for One-Color Microarray-Based Gene Expression Analysis (version 5.0.1). Labeled probes were hybridized to custom-designed Agilent barley 44K microarrays, and raw data were generated using an Agilent microarray scanner and feature extraction software.

For confirmation of array data, total RNA isolation and cDNA synthesis were performed as described above using the same root material as for the array experiments. Aliquots of 10–20 ng cDNA were used as the template for quantitative PCR using primers specific for individual genes (Table S2), and constitutively expressed ubiquitin served as the internal standard.

Data analysis was performed using Bioconductor/R ( The Limma package (Smyth, 2004) of Bioconductor was used for expression analysis of differentially regulated genes. Therefore, data were read by read.maimage, filtered by flags, and normalized using quantile normalization of background-corrected log2-converted intensities (normalizeQuantiles). Using lmFit (Linear Model for Series of Arrays), a linear model was fitted to the log2 expression data for each probe, and (Compute Contrasts from Linear Model Fit) was used to obtain coefficients and standard errors for contrasts of the coefficients of the original model. An empirical Bayes method, ebayes (Empirical Bayes Statistics for Differential Expression), was used to calculate the moderatedt statistics. A table of the top-ranked genes from the linear model fit was extracted using topTable (Table of Top Genes from Linear Model Fit). Genes with a P value ≤0.05 that were at least twofold regulated at one time point were filtered and displayed by heatmap. The data discussed here have been deposited in NCBI’s Gene Expression Omnibus database (Edgar et al., 2002), and are accessible through GEO Series accession number GSE13756 (

Gene expression analyses

For elucidation of candidate gene expression in GA mutants, RNA was extracted, reverse-transcribed to cDNA, and used to analyse the expression patterns of defence-related genes (Table S2) using standard quantitative PCR protocols. Data were analysed using the statistical procedure ‘lm’ (linear model) as described above.

Determination of carbohydrates and free amino acids

Frozen samples were extracted, and glucose, fructose and sucrose were quantified using a coupled optical test at 340 nm as described by Stitt et al. (1989) in a total assay volume of 200 μl using a microtiter plate reader (BioTek,

Amino acids were derivatized using the fluorophore 6-aminoquinolyl-N-hydroxysuccimidyl carbamate (AccQ Tag, and subsequently resolved by HPLC analysis using a reverse-phase column (Luna C18; particle size 5 μm, length 250 mm, internal diameter 4.6 mm; Phenomex, as described by van Wandelen and Cohen (1997) with modifications as described by Abbasi et al. (2007).

Quantitation of carotenoids

Frozen root tissue (50 mg) was ground to a fine powder and extracted with 400 μl methanol by homogenization. During the following steps, samples were shielded from light and kept on ice. After incubation for 5 min at 4°C, 400 μl 50 mm Tris/HCl pH 8.0/1 m NaCl were added, and the mixture was incubated for 5 min before addition of 800 μl chloroform for extraction of the carotenoids from the methanol phase. Samples were inverted for 5 min, incubated for 10 min, and then centrifuged at 13 000 rpm for 5 min to achieve phase separation. Extraction with chloroform was repeated, and the lower phases were pooled and vacuum dried in a speed-vac.

For reverse-phase chromatography, extracts were dissolved in 200 μl eluent B (methanol:acetonitrile:isopropanol:water, 73:20:5:2), and 20 μl aliquots were resolved on a Dionex Acclaim PA C16 column (internal diameter 4.6 mm, length 150 mm, particle size 5 μm) using a Dionex Ultimate 3000 HPLC system ( connected to an ICS 2600 photodiode array detector (Hamamatsu, with a gradient as described by Fraser et al. (2000). Pigments were detected and quantified based on their absorption at 450 nm, and identified according to their specific 3D spectra between 300 and 700 nm.


This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG Research Group FOR666) to K.H.K., P.S. and U.S.