Biotrophy-specific downregulation of siderophore biosynthesis in C olletotrichum graminicola is required for modulation of immune responses of maize

Abstract

The hemibiotrophic maize pathogen C olletotrichum graminicola synthesizes one intracellular and three secreted siderophores. eGFP fusions with the key siderophore biosynthesis gene, SID1, encoding l-ornithine-N ⁵-monooxygenase, suggested that siderophore biosynthesis is rigorously downregulated specifically during biotrophic development. In order to investigate the role of siderophores during vegetative development and pathogenesis, SID1, which is required for synthesis of all siderophores, and the non-ribosomal peptide synthetase gene NPS6, synthesizing secreted siderophores, were deleted. Mutant analyses revealed that siderophores are required for vegetative growth under iron-limiting conditions, conidiation, ROS tolerance, and cell wall integrity. Δsid1 and Δnps6 mutants were hampered in formation of melanized appressoria and impaired in virulence. In agreement with biotrophy-specific downregulation of siderophore biosynthesis, Δsid1 and Δnps6 strains were not affected in biotrophic development, but spread of necrotrophic hyphae was reduced. To address the question why siderophore biosynthesis is specifically downregulated in biotrophic hyphae, maize leaves were infiltrated with siderophores. Siderophore infiltration alone did not induce defence responses, but formation of biotrophic hyphae in siderophore-infiltrated leaves caused dramatically increased ROS formation and transcriptional activation of genes encoding defence-related peroxidases and PR proteins. These data suggest that fungal siderophores modulate the plant immune system.

Introduction

Iron is an essential nutrient and the major redox element in all eukaryotes. Paradoxically, although iron is one of the most abundant elements on Earth, its bioavailability is extremely low. At neutral pH, the solubility product of Fe^(III), the main oxidation state in an aerobic environment, is less than 10⁻¹⁸ M (Neilands et al., ¹⁹⁸⁷). As compared to iron-rich terrestrial environments, the availability of iron may even be significantly lower for pathogenic fungi, as free iron does basically not exist in mammalian and plant hosts (Weinberg, 1978; 2009; Expert, ¹⁹⁹⁹), due to tight sequestration by ferritin, transferrins and/or several other iron-binding proteins (Ratledge and Dover, 2000; Arosio and Levi, ²⁰⁰²). As a consequence, pathogenic fungi have evolved two high-affinity iron acquisition strategies allowing them to compete with their hosts for iron, by either siderophore-mediated or reductive iron acquisition (Philpott, 2006; Haas et al., ²⁰⁰⁸).

All siderophores identified in pathogenic asco- and basidiomycetes so far belong to the hydroxamate type of siderophores (Haas et al., ²⁰⁰⁸). The first step in the biosynthesis of hydroxamate siderophores in fungi is N⁵-hydroxylation of l-ornithine, catalysed by the key enzyme of this pathway, l-ornithine-N⁵-monooxygenase (Haas et al., ²⁰⁰⁸; Fig. S1). Deletion of the corresponding gene, SID1, blocks synthesis of all intracellular and secreted siderophores. The Aspergillus nidulans mutant deficient in the SID1 homologue sidA exhibits severe growth defects, due to the lack of reductive iron assimilation (Eisendle et al., ²⁰⁰³). Closely related pathogenic Aspergillus species, i.e. the plant pathogen A. oryzae and the mammalian pathogen A. fumigatus, are able to utilize reductive iron assimilation, which compensates for the loss of siderophores. Therefore, the lack of siderophore biosynthesis affects vegetative growth under iron-limiting conditions of these pathogenic species only to some extent (Haas et al., ²⁰⁰⁸).

In subsequent steps of siderophore biosynthesis, the hydroxamate group is formed by N⁵-acylation of N⁵-hydroxyornithine, and siderophore synthesis is completed by covalently linking the hydroxamate residues by non-ribosomal peptide synthetases (NRPSs). NRPSs are large multifunctional enzymes with modular structure that synthesize peptides in a ribosome-independent manner (Mootz et al., ²⁰⁰²). These enzymes, which are exclusively found in bacteria and fungi, require activation by 4′-phosphopantetheinylation (Horbach et al., ²⁰⁰⁹; Marquez-Fernandez et al., ²⁰⁰⁷; Wiemann et al., ²⁰¹²; Fig. S1). Only one of the NRPSs, i.e. Nps6, is required for synthesis of secreted siderophores. In A. fumigatus and in several necrotrophic plant pathogens, i.e. the Southern Corn Leaf Blight fungus Cochliobolus heterostrophus; the causal agent of brown spot disease in rice, Cochliobolus miyabeanus; the causal agent of head blight of wheat and barley and ear rot of maize, Fusarium graminearum; and the causal agent of black spot disease on virtually all economically important Brassica species, Alternaria brassicicola, inactivation of NPS6 homologues affected rates of vegetative growth, asexual sporulation, and/or oxidative stress tolerance under iron-limiting conditions (Oide et al., ²⁰⁰⁶; Schrettl et al., ²⁰⁰⁷). Intracellular siderophores are synthesized by Nps2 (Haas et al., ²⁰⁰⁸).

Competitive iron acquisition is indispensable for hyphal growth in iron-limited environments and can therefore be regarded as a pathogenicity factor in both mammalian and plant hosts. Deletion of the l-ornithine-N⁵-monooxygenase gene sidA of the mammalian pathogen A. fumigatus abrogated siderophore biosynthesis and prevented the establishment of pulmonary aspergillosis in a mouse model (Schrettl et al., ²⁰⁰⁴; Hissen et al., ²⁰⁰⁵). In contrast, inactivation of reductive iron assimilation in this fungus had no effect on virulence (Schrettl et al., ²⁰⁰⁴). Likewise, in the above mentioned necrotrophs F. graminearum, C. heterostrophus, C. miyabeanus and A. brassicicola, siderophore-mediated iron uptake plays the predominant role in virulence (Oide et al., ²⁰⁰⁶; Park et al., ²⁰⁰⁶; Greenshields et al., ²⁰⁰⁷). NPS2 deletion experiments in the rice blast fungus Magnaporthe oryzae suggested that the intracellular siderophore ferricrocin may also be involved in virulence (Hof et al., ²⁰⁰⁷).

By contrast, in the biotrophic plant pathogenic smut fungi Microbotryum violaceum, infecting white campion (Silene latifolia), and Ustilago maydis, infecting maize (Zea mays), reductive iron assimilation, but not siderophore-mediated iron uptake, is required for full virulence (Mei et al., ¹⁹⁹³; Birch and Ruddat, ²⁰⁰⁵). Supporting the idea that reductive iron assimilation is required for virulence of biotrophs, Albarouki and Deising (2013) have shown that the ferroxidase gene FET3-1 of the hemibiotroph Colletotrichum graminicola is specifically expressed in and required for development of biotrophic hyphae, and that this gene is a determinant of virulence.

Hemibiotrophic plant pathogens like C. graminicola are ideally suited to analyse the role of the two fungal high-affinity iron acquisition strategies, mediated via either siderophores or reductive iron assimilation, in the same genetic background during pathogenesis. As a first step in the infection process, the maize leaf anthracnose and stem rot fungus C. graminicola differentiates a dome-shaped infection cell, called an appressorium, on the epidermis of the leaf. The invading infection vesicle and primary hyphae invaginate the plasma membrane of the host and re-programme the host cell, resulting in prevention of host cell death and re-direction of nutrients to the infection site (Behr et al., ²⁰¹⁰). During this biotrophic stage of pathogenesis, the hyphae are tightly surrounded by the plasma membrane of the plant, and the bifacial matrix surrounding the hyphae is likely extremely poor in nutrients, including iron. Upon perception of a hitherto unidentified signal, the pathogen changes its lifestyle and differentiates thin, fast growing, highly destructive necrotrophic hyphae (Bergstrom and Nicholson, 1999; Horbach et al., ²⁰¹¹; Horbach and Deising, ²⁰¹³). However, whether employment of the two high-affinity iron acquisition strategies is functionally linked to the development of biotrophic and necrotrophic infection structures, and whether differential utilization of these strategies is required to establish a hemibiotrophic lifestyle and a compatible parasitic interaction is unknown.

Using SID1:eGFP fusions and live-cell imaging, we show specific expression of SID1 in secondary hyphae, indicating that siderophore-mediated iron acquisition is employed during necrotrophy, and rigorously downregulated in biotrophic hyphae. Targeted deletion of SID1 and NPS6 demonstrated that siderophore biosynthesis is required for efficient spreading of necrotrophic hyphae and full virulence of C. graminicola. Furthermore, diaminobenzidine staining and qRT-PCR showed that leaves infiltrated with the siderophore coprogen activated broad spectrum defence responses when biotrophic hyphae invaded epidermal cells. Siderophore infiltration without subsequent inoculation did not evoke defence responses.

These data suggest that secretion of siderophores during infection by biotrophs modulates the plant immune system.

Results

The availability of iron controls expression of the siderophore biosynthesis genes SID1, NPS2 and NPS6

Genome mining of pathogenic Pezizomycotina, Taphrinomycotina and some Basidiomycota suggested that remarkable differences in fungal iron acquisition exist, possibly depending on their lifestyle. Most filamentous Ascomycota, e.g. several Aspergillus, Fusarium and Mycosphaerella species, harbour only a single copy of the key gene of siderophore synthesis, SID1, and thus, a single l-ornithine-N⁵-monooxygenase (Fig. S2A; Sid1). This also holds true for the maize anthracnose fungus C. graminicola. The biotrophic maize pathogen and basidiomycete U. maydis harbours a SID1 gene, but utilizes reductive iron assimilation during pathogenesis (Eichhorn et al., ²⁰⁰⁶). The rust fungus Puccinia graminis, also a biotroph, has a SID1-related gene, but the identity of the derived protein with that of SID1 of C. graminicola is only 25%, so that functional assays would be required to confirm that the rust protein indeed exhibits l-ornithine-N⁵-monooxygenase activity. Interestingly, the genome of the biotrophic powdery mildew fungus B. graminis f. sp. hordei does not contain a SID1 homologue.

In contrast to Sid1 proteins, a multitude of NRPSs exists in fungal species, as derived from the genomes inspected. In C. graminicola, nine NRPS proteins, and at least seven NRPS/PKS hybrids exist. Nps6 proteins are highly conserved among the Euascomycota, including plant and human pathogens (Fig. S2B). Functional analysis of all non-ribosomal peptide synthetases in C. heterostrophus revealed that only Nps6, the enzyme required to synthesize secreted siderophores, is involved in virulence and resistance to oxidative stress (Lee et al., ²⁰⁰⁵; Oide et al., ²⁰⁰⁶). Only a single copy of NPS6 exists in the genome of C. graminicola (Fig. S2B).

The availability of iron controls transcript abundance of the siderophore synthesis genes SID1, NPS2 and NPS6. Under iron-limiting conditions, as induced by addition of 200 μM of the iron scavenger bathophenanthroline disulphonate (BPS), enhanced levels of SID1, NPS2 and NPS6 transcripts were found in vegetative hyphae. Fe^(III), which is taken up by siderophores, efficiently downregulated SID1 and NPS6 transcripts at concentrations corresponding to 10 μM or 1 mM. Interestingly, high iron concentrations led to increased transcript levels of NPS2 required for synthesis of intracellular storage siderophores and control of intracellular iron homeostasis. Similar transcriptional responses were observed after application of Fe^(II), although adjusting the Fe^(II) concentration to 10 μM reduced SID1 transcript abundance only slightly (Fig. 1A). To use an independent method showing that iron availability regulates siderophore biosynthesis, a C. graminicola strain harbouring an eGFP gene fused to the SID1 promoter was constructed (Fig. S3), and eGFP fluorescence of vegetative hyphae was used as a measure of SID1 promoter activity. As expected, in the presence of 200 μM BPS strong eGFP fluorescence was found, while in the presence of 1 mM FeSO₄ eGFP fluorescence decreased dramatically (Fig. 1B).

Figure 1.

Iron availability regulates transcript abundance of SID1, NPS2 and NPS6.A. Relative transcript abundance of the siderophore synthesis genes SID1, NPS2 and NPS6 at different iron supplies. Actin ACT1 and histone H3 transcripts were used as references.B. Control of SID1 expression by the availability of iron, as measured by the eGFP-fluorescence of the WT strain harbouring the P_SID1:eGFP fusion. Fluorescence intensity is given in relative units.

SID1 and NPS6 are required for vegetative growth under iron-limiting conditions, tolerance to ROS, and conidiation

In order to functionally characterize siderophore biosynthesis, we deleted the ornithine-N⁵-monooxygenase gene SID1, required for synthesis of all hydroxamate siderophores, as well as the non-ribosomal peptide synthetase gene NPS6, specifically required for synthesis of all secreted siderophores (Fig. S4). The growth defect under iron-limiting conditions, as well as the defects in infection structure differentiation and virulence were fully complemented by transforming the Δsid1 deletion strain with a SID1:eGFP construct. Furthermore, all independent deletion strains tested showed identical phenotypes, collectively indicating that the phenotype of the Δsid1 deletion strain was indeed caused by the deletion of SID1. Genetic complementation of Δnps6 mutants was not successful, due to the large size of the NPS6 gene. However the mutant phenotype was rescued by chemical complementation (see below), suggesting that the phenotype of all independent Δnps6 strains was indeed due to the deletion of NPS6. The WT strain CgM2 and the resulting Δsid1 and Δnps6 mutants of C. graminicola were grown in Sundström minimal medium inducing siderophore biosynthesis (Sundström, 1964), and the culture filtrates as well as the cell homogenates were used to analyse secreted and intracellular siderophores. HPLC (Fig. 2), in combination with mass spectrometry analyses (Fig. S5), revealed that C. graminicola produces three secreted siderophores, i.e. coprogen, coprogen B, and methyl-coprogen B, as well as the intracellular storage siderophore ferricrocin, confirming and extending previous data (Horbach et al., ²⁰⁰⁹). In addition, dimerumic acid was identified, representing either a siderophore precursor or a degradation product (Fig. 2 and Fig. S5). As expected, Δsid1 mutants were not able to synthesize any siderophore, whereas Δnps6 mutants were specifically devoid of secreted siderophores. Thus, HPLC-MS analyses fully confirmed the roles of SID1 and NPS6 in the siderophore biosynthesis pathway and showed that targeted deletion was successful.

Figure 2.

HPLC-MS analysis of secreted and intracellular siderophores of C . graminicola strains.A. Secreted siderophores analysed in culture media of the WT, Δsid1 and Δnps6, strains. Δsid1 and Δnps6 are unable to produce any extracellular siderophores produced by the wild-type strain (WT).B. Intracellular siderophore isolated from vegetative fungal mycelia. WT and Δnps6, but not Δsid1 strains are able to produce the intracellular siderophore ferricrocin.Peak 1, dimerumic acid; peak 2, coprogen B; peak 3, N^α-methylcoprogen B; peak 4, coprogen; peak 5, ferricrocin.

In several fungi, iron sufficiency is indispensable for asexual sporulation, vegetative growth and ROS tolerance (Oide et al., ²⁰⁰⁶). Although two high-affinity iron uptake systems, i.e. siderophore-mediated iron acquisition and reductive iron assimilation, exist, decreasing iron availability caused by increasing concentrations of BPS in the growth medium led to dramatically decreased conidiation rates of the WT strain (Fig. 3A). Conceivably, Δsid1 mutants were unable to produce conidia (Fig. 3B); strains harbouring an ectopically integrated SID1 deletion cassette were indistinguishable from the WT strain (Fig. S6). Addition of enterobactin, an archetype bacterial siderophore (Raymond and Carrano, 1979) did not restore conidiation, and supplementing oat meal agar (OMA) with a mix of mugineic acid-type phytosiderophores (Meda et al., ²⁰⁰⁷) or with rhizoferrin, the siderophore originally identified in Rhizopus microsporus (Drechsel et al., ¹⁹⁹¹), only partially restored conidiation. However, the intracellular siderophore ferricrocin as well as spent medium of C. graminicola, containing coprogen, coprogen B, and methyl-coprogen B, fully restored conidiation (Fig. 3B).

Figure 3.

Effect of iron limitation on conidiation of C . graminicola.A. The WT strain CgM2 showed decreased growth and conidiation rates at increasing BPS concentrations in OMA medium.B. Supplementing the growth medium with different types of siderophores partially or fully restored conidiation of the Δsid1 strain. The hydroxamate type siderophore ferricrocin or the filtrate of CgM2 spent medium fully complemented conidiation rates, whereas the bacterial siderophore enterobactin and phytosiderophores only partially restored conidiation.Different letters represent significant difference groups (P < 0.001).

BPS-induced iron deficiency strongly reduced growth rates of both siderophore mutants generated in this study, Δsid1 and Δnps6, as compared to the WT strain and strains ectopically harbouring the cassettes designed to delete the SID1 or NPS6 gene respectively (Fig. 4). Likewise, sensitivity to ROS, either H₂O₂ or superoxide, generated by addition of rose bengal to the growth medium and incubation of plates under light, was severely increased in both mutants (Fig. 4). As the strains with the ectopically integrated SID1 or NPS6 deletion cassettes did not differ from the WT with respect to growth rate under iron-deficient or -sufficient conditions as well as sensitivity to ROS, only the ectopic strain harbouring the SID1 deletion cassette was used side by side with the WT strain as a control in further experiments.

Figure 4.

Effect of iron limitation and reactive oxygen species (ROS) on growth rates of the WT, Δsid1 and Δnps6 strains of C . graminicola.A. Growth and colony phenotype of the WT strain and the Δsid1 mutant and its corresponding ectopic strain (ect. SID1), as well as of the Δnps6 mutant and its corresponding ectopic strain (ect. NPS6) on potato dextrose agar (PDA), PDA containing 100 μM BPS (BPS), 0.01% H₂O₂ (H₂O₂) or 100 μg ml⁻¹ of the singlet oxygen-generating agent rose bengal (RB). No differences between WT and the ectopic strains were observed, but Δsid1 and Δnps6 showed reduced growth under iron-limited condition and were hypersensitive to ROS.B. Quantitative growth assays show that both BPS and ROS significantly affect growth of Δsid1 and Δnps6 mutants. Different letters represent significance groups at P = 0.001. Bar = + SD, n = 4.The growth rates were measured and photos were taken 4 DPI (PDA, BPS and H₂O₂) and 13 DPI (RB).

Interestingly, Δnps6 strains showed no conidiation after 2 and sparse conidiation after 3 weeks of incubation. Supplementing the growth medium with the siderophore coprogen restored conidiation of Δnps6 strains (Fig. 5A). Increased magnification revealed that salmon-coloured asexual fruiting bodies called acervuli were formed by the WT, by Δnps6 strains, and by ectopic transformants (Fig. 5B, arrowheads).

Figure 5.

Phenotype of Δnps6 strains of C . graminicola.A. The C. graminicola Δnps6 strain showed delayed growth rates and reduced rates of formation of acervuli (compare Δnps6 frown for 2 and 3 weeks; white arrowheads indicate acervuli) and conidiation in comparison to the WT and ectopic strains (ect.). The addition of coprogen complemented formation of acervuli and conidiation in Δnps6 strains.B. In comparison to WT and ectopic strains, Δnps6 strains formed small acervuli (arrowhead).Bars are 200 μm.

In liquid media spores formed by Δsid1 and Δnps6 strains exhibited defective shapes (Fig. S7A). Conidia of Δsid1 and Δnps6 strains grown in liquid complete medium (CM) were significantly smaller than those produced by the WT strain and strains harbouring an ectopically integrated deletion cassette (Fig. S7B).

These data demonstrate that siderophore-mediated iron uptake is required for conidiation.

Siderophore biosynthesis genes are required for cell wall integrity, appressorial penetration and full virulence

Efficient competitive iron uptake by fungal hyphae invading a host plant is required for virulence (Haas et al., ²⁰⁰⁸; Albarouki and Deising, ²⁰¹³). To test whether siderophore-mediated iron uptake is also required for virulence of C. graminicola, we inoculated non-wounded and wounded maize leaf segments as well as intact plants with falcate or oval conidia of the WT strain and of Δsid1 and Δnps6 deletion strains. Both falcate and oval conidia are able to differentiate melanized appressoria and to infect plants through intact surfaces. All strains formed oval conidia in liquid media without additional supply of iron. On OMA plates, the mutants required the addition of 100 μM Fe^(III)-EDTA to form falcate conidia (Fig. S6). On non-wounded leaves, Δsid1 and Δnps6 strains were unable to cause major disease symptoms and were dramatically reduced in virulence (Fig. 6A). The WT strain and strains harbouring an ectopic copy of the SID1 deletion cassette caused strong symptoms. On wounded leaves, where the fungus omits formation of biotrophic infection structures and directly differentiates necrotrophic hyphae, the WT and the strain harbouring an ectopically integrated deletion cassette showed severe disease symptoms. In comparison with these strains, disease symptoms caused by Δsid1 and the Δnps6 strains were weaker. Mock-inoculated leaves did not show any symptoms (Fig. 6A). On leaves of intact plants, the WT strain and the strain harbouring an ectopically integrated deletion cassette, but not the Δsid1 and Δnps6 mutants, caused disease symptoms (Fig. 6A). These data were fully confirmed by qPCR analyses (Fig. 6B). Studying rates of formation of pre-penetration infection structures revealed that germination and appressorium differentiation rates, including melanization, were reduced in both Δsid1 and Δnps6 deletion strains. Strains carrying an ectopically integrated deletion cassette exhibited appressorium differentiation rates comparable to that of the WT strain (Fig. 6C). However, appressoria that were able to invade the plant differentiated secondary hyphae in planta (Fig. 6D). Although the secondary hyphae formed by Δsid1 and Δnps6 deletion strains were morphologically comparable to those formed by the WT and ectopic strains (Fig. 6D), they showed growth retardation (Fig. 6B). Therefore, reduced virulence of Δsid1 and Δnps6 strains is due to both, a penetration defect and reduced growth rates of spreading of necrotrophic secondary hyphae, strongly suggesting that siderophores are required for necrotrophic development.

Figure 6.

Virulence of the WT, ectopic, Δsid1 and Δnps6 strains.A. Virulence assay on non-wounded detached leaves (4 DPI) showed that WT and ectopic (ect.) strains, but not the representative Δsid1 and Δnps6 strains, caused the anthracnose symptoms. On wounded leaf segments, Δnps6 and Δsid1 were able to colonize the leaf tissue, but symptom severity was reduced, as compared with symptoms caused by WT and ectopic strains. On intact whole plants, Δsid1 and Δnps6 strains were unable to cause disease symptoms.B. Quantification of fungal mass in infected leaves by qPCR at 4 DPI. Error bars are standard deviations. Different letters represent different significance groups (P ≤ 0.001, n = 3).C. Differentiation of infection structures on the plant surface by the WT strain, an ectopic strain (ect.), and Δsid1 and Δnps6 strains at 24 HPI. White bars, germinated conidia; grey bars, non-melanized appressoria; black bars, melanized appressoria. Error bars are standard deviations. Different letters represent different significance groups (P ≤ 0.001, n = 1200).D. Microscopical analyses of the WT strain, an ectopic strain (ect.), and Δsid1 and Δnps6 strains were done at 48 HPI; arrows, appressoria; arrowheads, necrotrophic secondary hyphae. Bar = 10 μm.

While the WT strains and the strain harbouring an ectopically integrated deletion construct formed massively sporulating acervuli, the Δsid1 and Δnps6 strains only differentiated initials of acervuli and dramatically reduced spore numbers on both non-wounded and wounded leaves (Fig. S8).

Iron deficiency activates the cell wall integrity pathway in the mammalian pathogen A. fumigatus (Jain et al., ²⁰¹¹), and mutants of C. graminicola deficient in the ferroxidase FET3-1 and reductive iron assimilation exhibit dramatically reduced class V chitin synthase transcript abundance, severe appressorial cell wall defects and reduced penetration rates (Albarouki and Deising, 2013). In order to address the question as to whether siderophore mutants exhibit infection structure abnormalities, we investigated on planta differentiation of infection structures microscopically. Amazingly, while conidia (Fig. 7A, WT, white arrowhead) of the WT strain germinated and differentiated intact melanized appressoria (Fig. 7A, WT, black arrowheads), approximately 50% of the conidia (Fig. 7A, Δsid1, white arrows) and appressoria (Fig. 7A, Δnps6, black arrows) of the mutants lysed and released their cytoplasm onto the plant surface (Fig. 7A, Δsid1 and Δnps6, white arrows). Transformants with an ectopically integrated deletion cassette differentiated intact infection cells (data not shown).

Figure 7.

Δsid1 and Δnps6 mutants exhibit severe cell wall defects, hypersensitivity to cell wall-degrading enzymes and altered transcript abundances of genes involved in cell wall biogenesis.A. Conidia (white arrowheads) and appressoria (black arrowheads) of Δsid1 and Δnps6 mutants lysed and released cytoplasm (black arrowhead) onto maize surface 24 h after inoculation, whereas the WT strain formed melanized dome-shaped appressoria (ap). Bars = 10 μm.B. Protoplasting efficiency assay show that oval conidia of Δsid1 (black triangles), Δnps6 (black circles), strains released protoplasts much faster than those of the WT (black squares) and the ectopic strain (white squares). Three individual replicates were counted of each strain, with 100 conidia each at each time point. Bars represent ± SD.C. Transcript abundance of the chitin synthase genes CHSI, CHSIII and CHSV and the β-1,3-glucan synthase gene GLS1 in appressoria of WT and Δsid1 strains differentiated on polypropylene sheets. Asterisks indicate significant difference between WT and Δsid1 (P ≤ 0.001). Bars represent + SD, n = 3.

In order to confirm cell wall defects in both Δsid1 and Δnps6 mutants, we performed protoplasting experiments with oval conidia collected from liquid growth medium (Fig. 7B). In both mutants, approximately 40% of the conidia had protoplasted one h after incubation in osmotically stabilized solution containing lysing enzymes. After three h, more than 80% of the conidia of the mutants, but only ∼ 8% of the WT conidia and ∼ 13% of the conidia of the ectopic strain had protoplasted (Fig. 7B), clearly indicating cell wall defects of the Δsid1 and Δnps6 mutants.

In Δsid1 mutants, which do not synthesize any siderophores, transcript abundances of genes encoding enzymes forming structural cell wall polymers, i.e. the three chitin synthase genes CHS1, CHSIII and CHSV, as well as the β-1,3-glucan synthase gene GLS1, were quantified. Transcript concentrations of all of these genes were severely reduced in Δsid1 mutants, as compared to the WT strain (Fig. 7C).

Collectively, these data show that siderophore-mediated iron uptake is required for cell wall integrity of pre-penetration differentiated infection structures and for full virulence.

Biotrophy-specific downregulation of biosynthesis and uptake of siderophores is required for modulation of plant defence responses

Comparison of high-affinity iron acquisition pathways in pathogenic fungi differing in lifestyle suggested that biotrophs utilize reductive iron assimilation, whereas necrotrophs employ siderophore mediated iron uptake (Haas et al., ²⁰⁰⁸). We performed infection cell-specific gene expression studies, taking advantage of transformants harbouring the eGFP fusions with the SID1 (SID1:eGFP) as well as with the SIT1 gene (SIT1:eGFP). Out of four predicted siderophore uptake transporters present in the genome of C. graminicola, SIT1 is the most strongly transcribed (O’Connell et al., ²⁰¹²). eGFP tagging is required to investigate infection cell-specific regulation of expression of individual genes, as differentiation of infection structures of this fungus does not occur in a sufficiently synchronized manner to employ qRT-PCR. Appressoria of SID1:eGFP strains fluoresced strongly on the maize cuticle [Fig. 8A, 24 hours post inoculation (HPI), ap]. Surprisingly, immediately at penetration of the plant cell wall eGFP fluorescence disappeared completely in biotrophic primary hyphae (Fig. 8A, 36 HPI, ph). At later stages of pathogenesis, i.e. when the fungus had switched to destructive necrotrophic development, intensive eGFP fluorescence re-appeared in secondary hyphae (Fig. 8A, 72 HPI, sh). Strong eGFP fluorescence was also observed in secondary hyphae at 96 HPI and in acervuli initials (Fig. 8A, 96 HPI, sh and ai), as well as in conidia newly formed in the acervulus (Fig. 8A, 120 HPI, co and av).

Figure 8.

Infection structure-specific expression of the siderophore synthesis gene SID1 and the siderophore transporter gene SIT1A. Expression of SID1 was assessed in a C. graminicola replacement strain harbouring a SID1:eGFP fusion construct. The SID1 expression is downregulated during the biotrophic phase (36 HPI) and occurred again during the following necrotrophic phase.B. Expression of SIT1 as assessed in a SIT1:eGFP strain. The expression of the SIT1 gene perfectly matches the expression of SID1 in A.C. Expression of FET3-1 was assessed in a C. graminicola replacement strain harbouring a FET3-1:eGFP fusion construct. FET3-1 is expressed in infection structures formed on the plant surface and during biotrophy.ap, appressorium; co, conidium; ph, primary hypha; sh, secondary hypha; ai, acervulus initial; av, acervulus. iv, infection vesicle; st, sterigma. Bars, 24–72 HPI = 10 μm, 96–120 HPI = 50 μm.

In SIT1:eGFP strains, conidia and germ tubes fluoresced, but no fluorescence was observed in appressoria (Fig. 8B, 24 HPI; co, gt, ap). As in the SID1:eGFP strain, no fluorescence was seen in biotrophic infection vesicles and in primary hyphae (Fig. 8B, 36 HPI, iv, ph), but fluorescence re-occurred in secondary hyphae (Fig. 8B, 72–120 HPI, sh). Sterigmata did neither fluoresce in SID1:eGFP nor in SIT1:eGFP strains (Fig. 8A and B, 120 HPI, st). In contrast, FET3-1:eGFP strains showed strong fluorescence of biotrophic infection structures (Fig. 8C, 36 HPI, ph), and fluorescence rapidly disappeared when necrotrophic secondary hyphae formed (Fig. 8C, 72 HPI, sh). Importantly, virulence of all eGFP strains used did not differ from that of the WT strain (Fig. S3H, and Albarouki and Deising, 2013).

Cell type-specific expression of SID1, SIT1 and FET3-1 clearly visualized complementary employment and lifestyle specificity of the two high-affinity iron uptake strategies. Based on these data we hypothesized that siderophore secretion during biotrophic development may impose a significant disadvantage to the pathogen. Bacterial siderophores have been shown to activate transcription of ferritin genes, and as sequestration of iron by these proteins led to reduced pathogen growth, activation of ferritin genes has been regarded as a plant defence response (Expert, 1999; Dellagi et al., ²⁰⁰⁹; Kieu et al., ²⁰¹²). Greenshields et al. (2007) hypothesized that defence responses in plants may also be modulated by fungal siderophores. As simultaneous overexpression of all genes involved in the siderophore biosynthetic and secretory pathway in C. graminicola is difficult to accomplish, we decided to infiltrate coprogen, one of the siderophores secreted by this fungus, into maize leaves, and to subsequently inoculate the infiltrated leaves with fungal conidia.

Generation of H₂O₂ is among the fastest defence responses initiated after perception of pathogen ingress. To investigate whether an H₂O₂ response was induced upon infiltration of maize leaves with coprogen, we visualized H₂O₂ by 3,3′diaminobenzidine (DAB) staining (De Neergaard, 2005). In order to discriminate between defence responses induced by coprogen and iron scavenging, leaves were infiltrated with both desferri-coprogen and iron-loaded coprogen (Fe-coprogen) as well as with EDTA as a further control. To test for responses induced by the infiltration process, infiltration was also performed with sterile distilled water. In non-inoculated leaves, no H₂O₂ formation was detected, irrespective of whether leaves were non-infiltrated, water-, EDTA- or siderophore-infiltrated (Fig. 9A). However, at 48 HPI with C. graminicola, i.e. when the fungus had established biotrophic hyphae, a dramatic H₂O₂ response was specifically observed in coprogen-pretreated leaves, independent of whether desferri- or Fe-coprogen had been used to infiltrate leaves (Fig. 9A). Importantly, only the coprogen-infiltrated leaf area attacked by melanized appressoria (Fig. 9A, middle panel) showed a DAB-positive response, whereas DAB staining at the margins of the inoculation site was negligible. Under supply of desferri- and Fe-coprogen, most host cells decorated with a fungal appressorium (Fig. 9A, white asterisk) showed strong DAB staining (Fig. 9A, black asterisk) as well as intensive cell wall staining (Fig. 9A, arrowheads). None of the other treatments tested elicited a H₂O₂ response (Fig. 9A). These observations suggest that coprogen, but not iron scavenging or mechanical stress exerted by the infiltration process, pre-conditioned the leaf to respond to subsequent infection by C. graminicola.

Figure 9.

Coprogen modulates plant defence responses.A. DAB staining of non-inoculated and inoculated leaves infiltrated with H₂O, EDTA, desferri-coprogen or iron-loaded coprogen (Fe-coprogen). Untreated leaves served as controls. Bars = 100 μm (upper and middle panel) and 10 μm (lower panel).B. Transcript abundance of genes encoding the maize peroxidases PRX346 (GRMZM2G093346); PRX365 (GRMZM2G117365); and PRX648 (GRMZM2G144648) of water-, EDTA- and coprogen-infiltrated leaves at 48 HPI with the WT strain. Error bars are + standard deviations; n = 3.C and D. Transcript abundance of maize defence marker genes encoding pathogenesis-related protein 1 (PR1) and chitinase I after 48 HPI; n.i., non-infiltrated; H₂O, water-infiltrated; coprogen, coprogen-infiltrated; −, non-inoculated; +, inoculated with the C. graminicola WT strain. Error bars are + standard deviations; n = 3.

Furthermore, also on the transcript level a strong modulation of defence responses by coprogen was identified. Infiltration of leaves with desferri-coprogen, followed by C. graminicola infection, did not only considerably upregulate the transcript abundance of three genes encoding cell wall-localized peroxidase, which use H₂O₂ as a co-substrate, e.g. for lignification of cell walls (Fig. 9B), but also of genes encoding classical PR proteins such as PR1 and a secreted type-I chitinase (Figs. 9C and D). Peroxidase genes were also activated to some extent in EDTA-infiltrated leaves (Fig. 9B).

Taken together, these data suggest that the siderophore coprogen acts as an immunomodulating compound, and that the hemibiotroph C. graminicola downregulates biosynthesis and secretion of siderophores during biotrophic development in order to evade plant immune responses such as H₂O₂ formation, cell wall reinforcement and activation of PR gene expression.

Discussion

Synthesis of immunomodulating siderophores is downregulated in biotrophic fungal infection structures to avoid initiation of defence responses in maize

Biotrophy-specific expression of genes of C. graminicola involved in reductive iron assimilation (Albarouki and Deising, 2013) raised the question whether either one of the two iron acquisition strategies bears advantages or disadvantages at certain stages of pathogenesis. In this article we show that the expression of the key gene of siderophore biosynthesis, SID1, as well as of the siderophore transporter SIT1 is rigorously downregulated during biotrophic development of the maize anthracnose fungus C. graminicola, but strongly expressed during necrotrophy (Fig. 8). Restriction of siderophore-mediated iron uptake to necrotrophic infection structures brought up the hypothesis that secretion of siderophores into the narrow interfacial matrix surrounding biotrophic infection structures may impose a significant disadvantage to this pathogen during biotrophic development.

An adverse role of siderophores in plant infection was suggested by studies involving bacterial plant pathogens, which showed that siderophores may indirectly immunomodulate plants. In Arabidopsis thaliana, the gene FER1 encoding the iron-binding protein ferritin, is massively activated after infection with the WT strain of the enterobacterium Erwinia chrysanthemi. This bacterium secretes the monocatecholate-type siderophore chrysobactin as well as the citrate-derived siderophore achromobactin (Expert, 2005; Boughammoura et al., ²⁰⁰⁷). In contrast, chrysobactin- and achromobactin-deficient mutants of E. chrysanthemi did not transcriptionally activate FER1. Furthermore, significant activation of FER1 occurred in Arabidopsis leaves infiltrated with 100 μM chrysobactin, but not in leaves infiltrated with the iron-loaded form of this siderophore, and similar responses to the siderophore desferrioxamine and to its ferric complex have been observed (Dellagi et al., ²⁰⁰⁵). Since in these experiments alternative iron-complexing agents such as EDTA were not tested, it remained unclear to what extent siderophor-mediated ferritin regulation reflects a consequence of altered cytosolic iron availabilities in the host cell or whether siderophores may directly induce defence responses.

Our present study suggests that a fungal siderophore acts as an immunomodulating agent. Infiltration of maize leaves with coprogen, one of the secreted siderophores of C. graminicola, without subsequent inoculation, did neither induce formation of H₂O₂, nor induce transcription of genes encoding pathogenesis-related proteins such as PR1 and chitinases I, which are involved in broad spectrum disease resistance. However, when coprogen-infiltrated leaves were inoculated with C. graminicola, massive DAB staining was observed at 48 HPI, i.e. when the fungus had established biotrophic hyphae, and the H₂O₂ response was spatially associated with those cells attacked by the pathogen. This H₂O₂ response did not only occur in leaves infiltrated with desferri-coprogen, but also in leaves infiltrated with the iron-loaded siderophore (Fig. 9A). Neither non-infiltrated control leaves, nor leaves infiltrated with H₂O or leaves infiltrated with the iron-complexing agent EDTA showed an H₂O₂ response, irrespective of whether or not leaves had been inoculated. As H₂O₂ is a co-substrate for peroxidase-mediated cell wall re-enforcement, we analysed the transcript abundance of three cell wall-localized peroxidase genes of maize. Transcript abundance was clearly increased in coprogen pre-infiltrated infected leaves, as compared with infected leaves that had previously been infiltrated with H₂O (Fig. 9B). Moreover, inoculation with C. graminicola significantly increased PR1 and class I chitinase transcript levels in coprogen-infiltrated, but not in H₂O infiltrated or non-infiltrated control leaves (Fig. 9C and D). Collectively, these data strongly suggest that the siderophore coprogen modulates broad spectrum defence responses of maize and can therefore be regarded as a priming agent (Conrath, 2011). Interestingly, in various plants, including barley (Hordeum vulgare), bean (Phaseolus vulgaris) and potato (Solanum tuberosum) priming was accomplished by applying spent culture filtrate of Bacillus subtilis (B50) but not by applying culture fluid harvested during the logarithmic growth phase. Components in the spent culture filtrate bound iron, suggesting that bacterial siderophores were the active compounds. Indeed, spent media contain high concentrations of siderophores, as iron is quickly consumed and represents a limiting factor for microbial growth (Letendre and Gibbons, 1985). Interestingly, induced resistance was only active against biotrophs such as the barley powdery mildew and the bean rust fungus, and against Phytophthora infestans during biotrophic stages of pathogenesis (von Alten and Schönbeck, ¹⁹⁸⁶; U. Steiner-Stenzel, INRES, University of Bonn, Germany, pers. comm.).

Biotrophy-specific downregulation of siderophore biosynthesis appears to be required for evasion of defence responses. Comparably, in C. graminicola biotrophy-specific downregulation of formation of the structural cell wall component β-1,3-glucan, representing a PAMP, is required for evading defence responses in maize and for establishing a compatible parasitic interaction (Oliveira-Garcia and Deising, 2013). Thus, in this fungus the expression of genes, the products of which may contribute to recognition of the pathogen appear to be restricted to infection structures not developing in intimate contact with the plant plasma membrane.

Siderophore-mediated iron uptake is indispensable for cell wall integrity, asexual sporulation, and hyphal growth

In C. graminicola, like in the vast majority of pathogens analysed so far, high-affinity iron uptake systems are indispensable not only for hyphal growth under iron-limited conditions, but also for asexual sporulation (Eichhorn et al., ²⁰⁰⁶; Oide et al., ²⁰⁰⁶; Schrettl et al., ²⁰⁰⁷) (Figs 3 and 5). Indeed, in plant pathogenic fungi sporulation is a key requirement for spreading of diseases caused by air-borne pathogens such as rust fungi and powdery mildews (Limpert et al., ¹⁹⁹⁹; Hovmoller et al., ²⁰⁰⁸; Singh et al., ²⁰¹¹). Although long-distance dispersal of spores by water is thought to be unlikely (Bärlocher, 2009), spores produced in acervuli, which are embedded into a mucilaginous matrix and normally are dispersed by rain events, can be wind-dispersed over large distances on dried fragments of their host leaves (Nicholson and Moraes, 1980; Bergstrom and Nicholson, ¹⁹⁹⁹).

In order to identify genes required for conidiation, early work in A. nidulans employed a subtractive hybridization procedure called cascade hybridization to purify cDNAs complementary to poly(A) + RNA sequences. The results suggested that ∼ 1000 mRNAs are present in conidiating cultures, but not in vegetative hyphae (Timberlake, 1980; 1990). More recent work analysed transcriptional changes in the transition from vegetative cells to asexual sporulation, taking advantage of Illumina-based RNA sequencing. Interestingly, ∼ 20% of the genes were significantly altered at the transcriptional level, with 2035 genes being downregulated, and only 187 upregulated. Among the top 20 upregulated genes were genes encoding proteins which are involved in cell wall biogenesis, i.e. the putative β-1,3-transglycosidase gelD, a putative endo-mannanase, and two hydrophobins, including rodA (Garzia et al., ²⁰¹³). Notably, although not among the top 20 upregulated genes, the SID1 orthologue SidA was significantly upregulated during asexual sporulation (Garzia et al., ²⁰¹³). In line, in A. fumigatus the intracellular siderophore ferricrocin has been shown to be involved in intracellular iron trafficking, which is particularly crucial for conidiation (Wallner et al., ²⁰⁰⁹). Under iron deficiency, genes required for purine biosynthesis are downregulated, and although no known iron-dependent enzymes have been identified in this pathway in Saccharomyces cerevisiae, purine biosynthesis is an iron-dependent process in most organisms, including yeast and likely also filamentous fungi (Philpott and Protchenko, 2008). Therefore, iron limitation must have a drastic effect on transcription of genes required at specific stages of morphogenesis, and this likely holds true for genes involved in cell wall biogenesis in conidia and fungal infection structures (Garzia et al., ²⁰¹³; Oliveira-Garcia and Deising, ²⁰¹³).

Furthermore, in the human pathogen Candida albicans 11 genes encoding glycosylphosphatidylinositol (GPI)-anchored proteins, including RBT5, as well as several cell wall proteins, including BGL2, encoding a β-1,3-glucan transferase and PIR1, encoding a structural cell wall protein, are controlled by iron availability (Lan et al., ²⁰⁰⁴). The expression of the RBT5 gene of C. albicans, required for cell wall integrity, is regulated by the iron-responsive GATA-like factor Sfu1p (Mishra et al., ²⁰¹¹). Interestingly, also in C. graminicola cell wall integrity may be regulated by an iron-responsive GATA-like factor, as GATA binding sites are present at two positions (−163 and −431) of the 5′-noncoding region of the chitin synthase gene CHSV and at two positions (−17 and −282) of the 5′-noncoding region of the β-1,3-glucan synthase gene GLS1. Thus, control of cell wall biogenesis genes of C. graminicola might be administered by homologues of the GATA factor SreA, possibly in tight interplay with repressors such as Hap-like CCAAT binding complexes known from filamentous fungi such as A. fumigatus (Brakhage et al., ¹⁹⁹⁹; Hortschansky et al., ²⁰⁰⁷). The presence of putative Hap-like binding complexes at position −231, −481 and −490 in the 5′-noncoding regions of CHSI, CHSV and GLS1 genes, respectively, supports this assumption. Furthermore, in A. fumigatus the MAP kinase MpkA controls cell wall integrity, iron adaptation and oxidative stress response (Valiante et al., ²⁰⁰⁹; Jain et al., ²⁰¹¹). Accordingly, dramatic iron limitation in Δsid1 or Δnps6 mutants on the plant surface may lead to activation of MpkA, which, in turn, might activate a repressor of genes involved in cell wall biogenesis, eventually causing lysis of spores, germ tubes or appressoria differentiated on the plant cuticle (Fig 7A, Δsid1 and Δnps6) (Jain et al., ²⁰¹¹).

Indeed, in sid1 mutants of C. graminicola the transcript abundance of three different genes encoding chitin synthases, i.e. CHSI, CHSIII and CHSV, as well as of the β-1,3-glucan synthase gene GLS1 were clearly downregulated (Fig. 7C). As cell wall biogenesis is of particular importance during germination of conidia and rapid differentiation of infection structures on planta, transcription of genes and synthesis of enzymes involved in cell wall biogenesis may be particularly affected by limited availability of purine nucleotides. This may also explain why the addition of 0.5 mM BPS reduced conidiation by 98% (Fig. 3A). In both Δsid1 and Δnps6 mutants of C. graminicola cell wall defects are clearly visible in conidia and appressoria (Fig. 7A). Also in siderophore-defective mutants of F. graminearum and A. fumigatus conidiation is dramatically reduced (Schrettl et al., ²⁰⁰⁴; Oide et al., ²⁰⁰⁶).

Thus, as iron limitation causes significant reduction in abundance of fungal spores as well as failure in hyphal growth and infection structure differentiation, iron is not only a key factor for pathogenicity, but also for initiation and spreading of disease in the field.

Perspectives for employing siderophore-mediated iron uptake in plant protection

Our work and that of others has shown that genes encoding siderophore-synthesizing enzymes as well as genes encoding siderophore transporters are active under iron-limiting conditions, including development on the plant surface and/or in the host tissue (Fig. 8) (Jain et al., ²⁰¹¹). An excellent example of how siderophores can be used to increase the sensitivity of pathogens to chemical control agents is provided by sideromycins, which are natural siderophore drug conjugates produced by bacteria such as Actinomyces subtropicus, Streptomyces griseoflavus and Streptomyces violaceus. Chemical control of pathogenic bacteria is hampered by limited permeability of their outer membrane for antibiotics. An efficient strategy to overcome this membrane-mediated resistance is the misuse of bacterial iron transport systems by sideromycins (Möllmann et al., ²⁰⁰⁹). Already in 1951, the sideromycin albomycin has been discovered in cultures of A. subtropicus (Gause, 1955; Benz et al., ¹⁹⁸²), whereas S. violaceus produces the closely related sideromycin salmycin. Salmycin consists of an amino disaccharide linked to the ferrioxamine B-type siderophore danoxamine via a dicarboxylic spacer. Salmycins are thought to be taken up by hydroxamate transporters, and the antibiotic amino glycoside may be released inside the cell upon hydrolysis (Vértesy et al., ¹⁹⁹⁵). Alternatively, it is thought that iron reduction may trigger drug release via an intramolecular cyclization process (Roosenberg and Miller, 2000). Adopting the bacterial strategy of employing microbial high-affinity siderophore uptake systems to introduce natural siderophore drug conjugates, known as the Trojan Horse strategy, synthetic siderophore drug conjugates may be used to control pathogenic bacteria (Heinisch et al., ²⁰⁰²; Miller et al., ²⁰⁰⁹). To date, conjugates of siderophores with a variety of commercially available antibiotics with different cellular targets have been developed (Wencewicz et al., ²⁰⁰⁹). It is tempting to speculate whether corresponding conjugates of fungal siderophores with different fungicides, e.g. azoles, dicarboximides or strobilurins, could be used in a corresponding Trojan Horse strategy to combat plant pathogenic fungi.

Alternatively, enzymes involved in siderophore biosynthesis may be used as a target of novel antifungal compounds. For example, the bacterial pathogens Mycobacterium tuberculosis and Yersinia pestis, causing tuberculosis and plague in man, respectively, use the mycobactin and yersiniabactin siderophores, both of which have salicyl-capped non-ribosomal peptide-polyketide hybrid scaffolds. Small non-hydrolysable molecules such as acyl sulphamoyl adenosides mimic the substrate of the salicylate adenylating enzymes MbtA and YbtE and inhibit siderophore biosynthesis in a non-competitive fashion (Ferreras et al., ²⁰⁰⁵). Adopting this strategy, analogues of substrates of enzymes involved in fungal hydroxamate-type siderophore biosynthesis may be synthesized and tested for their antifungal activity and applicability in plant protection. Thus, the detailed understanding of fungal iron uptake strategies may help to develop novel strategies to combat pathogenic fungi and to improve plant protection.

Experimental procedures

Strains, media and culture condition

The wild-type strain CgM2 of C. graminicola was provided by R.L. Nicholson, Purdue University, Indiana. The CgM2 strain was grown on oat meal agar (OMA). The Δsid1 and Δnps6 were grown on OMA medium supplemented with 100 μM Fe^(III)-EDTA (Duchefa, Haarlem, the Netherlands). Alternatively, strains were grown in liquid complete medium (Leach et al., ¹⁹⁸²) in an incubation shaker (Unitron, Infors AG, Bottmingen, Switzerland) at 23°C and 100 r.p.m.

Sensitivity to reactive oxygen species and iron-limiting conditions were tested as described previously (Albarouki and Deising, 2013).

Infection assays

Fourteen-day-old maize (Z. mays cv. Nathan; KWS SAAT AG, Einbeck, Germany) plants were used in the infection assays. Quantitative PCR (qPCR) was employed for quantifying fungal mass of the infection sites as described previously (Albarouki and Deising, 2013).

Formation of conidia

The C. graminicola wild-type strain produces conidia when grown on OMA plates at 23°C and constant UV-light (TLD36W/08; Philips, Hamburg, Germany). The ability of the WT strain and the corresponding mutants to form conidia was assayed on OMA plates supplemented with different concentration of FeSO₄, FeCl₃, Fe^(III)-EDTA, siderophores (i.e. ferricrocin, coprogen, ferrichrome, rhizoferrin, phytosiderophores or enterobactin) or the iron scavenger BPS. Fifty-millimetre plates were filled with 10 ml OMA and inoculated with a 0.4-cm-diameter mycelial block using a cork borer from the front edge of the growing C. graminicola strains.

The phytosiderophores used in this study were collected from barley root exudates and purified according to Meda et al. (2007). Briefly, barley plants were first grown in iron-sufficient hydroponic solution for 2 weeks. After 1-week pre-culture under iron-deficiency, root exudates were collected daily and pooled. The root exudates were concentrated and prepurified by cation exchange chromatography (Dowex 50 WX column, Serva) and quantified by HPLC.

The plates were incubated at 23°C and constant UV light for 2 weeks. Thereafter, the conidia formed were washed from the plate with 2 ml distilled water, the washing process was repeated three times until the conidia were completely removed from the plate. The washed conidia were collected in 15 ml Falcon tube (Greiner bio-one, Frickenhausen, Germany), and counted in a haemocytometer (Thoma, LO-Laboroptik, Friedrichsdorf, Germany). Four plates were used per strain and growth assays were repeated four times.

Transcript abundance of genes involved in siderophore-mediated iron uptake

Mycelium of C. graminicola was grown in CM medium for 4 days. The grown mycelia were filtered through Miracloth (EMD Chemical, Darmstadt, Germany) and washed three times with sterile deionized water (Milli-Q, TKA, Thermo Electron LED GmbH, Niederelbert, Germany) under sterile conditions. Thereafter, 100 mg of the mycelium was transferred into iron free CM medium containing 50 μM of BPS for 2 days. Subsequently, the mycelium was transferred to fresh CM medium containing 0.01 mM or 1 mM ferric (FeCl₃), or ferrous (FeSO₄) iron, or 50 or 200 μM BPS for additional 2 days.

RNA isolation from C. graminicola was done according to Chirgwin et al. (1979). Genomic DNA was removed using Turbo DNA free ™ Kit (Life Technology, Darmstadt, Germany). Quantitative RT-PCR (qRT-PCR) was performed as described previously (Albarouki and Deising, 2013). All primers used for quantitative PCR are listed in Supplemental Table S1.

Targeted deletion of SID1 and NPS6 genes of C . graminicola

Deletion cassettes for the SID1 gene was generated by double-joint PCR (Yu et al., ²⁰⁰⁴). The 5′- and 3′-flanking regions of the SID1 gene were amplified with the primer combinations SID1-KO-5′-Fw and SID1-KO-5′-Rv, and SID1-KO-3′-Fw and SID1-KO-3′-Rv, and 100 ng of genomic DNA as template. The hygromycin phosphotransferase (hph) cassette was amplified from plasmid pAN7-1 (Punt et al., ¹⁹⁸⁷), using primers Hyg-Fw and Hyg-Rv. In a second PCR, the 5′- and 3′-flanks were fused with the hph cassette, and the complete product was amplified using nested primers SID1-KO-nest-Fw.1 and SID1-KO-nest-Rv.1. The 4519 bp product was cloned into pJET1.2/blunt, yielding pSID1KO. This vector was used as a template to amplify the KO construct using primers SID1-KO-nest-Fw.2 and SID1-KO-nest-Rv.2 primers and transformed into conidial protoplasts as described (Werner et al., ²⁰⁰⁷).

The inactivation of the NPS6 gene was achieved by a partial deletion of the ORF and substitution by hygromycin resistance cassette. First, a 5912 bp fragment of the NPS6 was amplified by PCR using primer pairs NPS6-Fw and NPS6-Rv, and cloned into pJET1.2/blunt. The resulting plasmid pNPS6 was digested with the AgeI endonuclease, resulting in the removal of 3229 bp of the ORF. The linearized plasmid was then dephosphorylated. The 2692 bp hygromycin resistance cassette was cut from pAN7-1 using (AgeI/NgoMIV) double digestion and subcloned into the AgeI-digested pNPS6, resulting pNPS6KO. The entire 5.6 Kb KO cassette was amplified from pNPS6KO by PCR using primer pairs NPS6-KO-nest-Fw and NPS6-KO-nest-Rv, and used to transform conidial protoplast.

All PCRs were performed using Phusion High-Fidelity DNA Polymerases (New England Biolabs, County Road, Ipswich, England).

Generation of P_SID1:eGFP, SID1 : eGFP and SIT1 : eGFP fusion constructs, and transformation into C . graminicola

The eGFP fragment was amplified by primer pairs EGFP-Fw and EGFP-Rv.1 from the plasmid pFET3-2-eGPF-G418 (Albarouki and Deising, 2013), resulting in a 1300 bp fragment. The SID1 promoter (P_SID1) was amplified by primer pairs SID1-Kpn2I-SfiIA-Fw and SID1-DraIII-SfiIB-Rv resulting in a 904 bp fragment. Both PCR fragments were purified and digested using DraIII. The two pieces were then ligated and amplified by PCR using primer pairs SID1-Kpn2I-SfiIA-Fw and EGFP-Rv.1. The resulting P_SID1:eGFP construct was then ligated into OliI-digested pNR1 plasmid (Malonek et al., ²⁰⁰⁴) resulting in the plasmid pEB14-eGFP.

The construction of the SID1:eGFP fusion cassette was mediated by the cloning of SID1 gene into plasmid pEB14-eGFP. The SID1 gene was amplified with primer pairs SID1-Kpn2I-SfiIA-Fw and SID1-SfiIB-Rv, and 100 ng of gDNA. The PCR fragment was digested with SfiI and cloned into plasmid pEB14-eGFP. The 5′-SfiI A and 3′-SfiI B overhangs allowed the oriented ligation of the SID1 gene in the frame of eGFP. The resulting plasmid pEB14-SID1-eGFP was sequenced and the ORF of the SID1:eGFP fused gene was confirmed.

To construct the SIT1:eGFP cassette, the SIT1 gene was amplified with SIT1-SfiIA-Fw and SIT1-SfiIB-Rv primers. The PCR product was digested with SfiI and cloned into the pEB14-eGFP plasmid in the frame with the eGFP gene. The resulting plasmid pEB14-SIT1-eGFP was confirmed by sequencing.

The P_SID1:eGFP, SID1:eGFP and SIT1:eGFP constructs were amplified by PCR, using primers pEB-uni-Fw and pEB-uni-Rv and transformed into conidial protoplasts as described (Werner et al., ²⁰⁰⁷).

All PCRs were performed using Phusion High-Fidelity DNA Polymerases (New England Biolabs, County Road, Ipswich, England).

Construction of the FET3-1:eGFP strain has been described previously (Albarouki and Deising, 2013).

DNA extraction and genomic Southern blot analyses

Extraction of genomic DNA from vegetative mycelium of C. graminicola was done as described (Döbbeling et al., ¹⁹⁹⁷). Five micrograms of genomic DNA was used for Southern blot analyses and digested with either MunI (SID1) or EcoRV (NPS6) and separated on a 0.7% (w/v) agarose gel as described before (Albarouki and Deising, 2013).

DIG-dUTP-labelled probes were generated by PCR, using primers Hyg-prob-Fw and SID1-KO-nest.Rv.2 and plasmid pSID1KO as template (SID1), or primers Hyg-prob-Fw and NPS6-KO-nest-Rv and plasmid pNPS6KO (NPS6). Taq polymerase was used for amplifying probes. Hybridization and probe detection was carried out as described (Roche Diagnostics, Mannheim, Germany), and membranes were exposed to Hyperfilm ECL X-ray film (Amersham Pharmacia Biotech Europe, Freiburg, Germany).

Siderophore analysis

Colletotrichum graminicola strains were grown in CM medium in an incubation shaker (Unitron, Infors AG, Bottmingen, Switzerland) at 25°C and 100 r.p.m. for 7–10 days. Then the fungal mycelia were filtrated under sterile conditions through Miracloth filter (EMD Chemical, Darmstadt, Germany) and washed three times with H₂O_bidest. The fungal mycelia were then dried gently with sterile filter paper (601A Carl Roth, Karlsruhe, Germany) and weighted. One gram fresh weight of fungal mycelium was used to inoculate 150 ml of Sundström minimal medium (Sundström, 1964) in 500 ml Erlenmeyer flasks. The inocula were grown at 25°C for an additional 10 days. The mycelia and the growth broth were separated by filtration (601A Carl Roth, Karlsruhe, Germany).

For analysis of intracellular siderophores, the mycelium was freeze dried. One hundred milligrams were homogenized in 1.8 ml of 0.05 M sodium-phosphate buffer, pH 6.5. One ml of the generated cell extract was iron saturated with 100 μl of 100 mM FeSO₄ and mixed thoroughly. After centrifugation to remove any precipitated iron, 10 μl were applied to reversed phase HPLC.

Extracellular siderophores were analysed from the supernatant obtained after a cultivation period of 10 days. Ten millilitres of each sample were applied to an Amberlite XAD-16 column (Rohm and Haas, Philadelphia, PA, USA) and equilibrated with water. Iron saturation of matrix-bound siderophores was achieved by addition of 100 μl of 100 mM FeSO₄. Siderophore-iron complexes were eluted with 2 ml of methanol and collected. Methanol was discarded by speed vacuum centrifugation overnight. The dried pellet was solubilized in 100 μl of water and 10 μl were applied to reversed phase HPLC.

Compounds displaying absorption at 430 nm, which is typical for iron-saturated siderophores, were subjected to mass spectrometry. Determination of the molecular mass of the samples obtained by RP-HPLC was carried out using a LTQ Velos ion trap mass spectrometer (Thermo Fisher Scientific, Vienna, Austria) equipped with an electrospray source (ESI-MS, electrospray ionization mass spectrometry, Vienna, Austria). Samples were dissolved in 50% aqueous methanol containing 0.1% formic acid, and infused directly into the ion source using the syringe pump. The electrospray voltage was set at 4.0 kV and the heated capillary was held at 270°C.

Protoplasting of oval conidia

Oval conidia formed in 0.5 M sucrose medium containing 0.1% yeast extract (Werner et al., ²⁰⁰⁷) were collected and suspended in a protoplast induction solution containing 20 mg ml⁻¹ lysing enzymes from Trichoderma harzianum (Sigma, Deisenhofen, Germany); 0.1% (v/v) β-mercaptoethanol in 0.7 M NaCl. The suspension was incubated at 30°C and 80 r.p.m. (Unitron, Infors AG, Bottmingen, Switzerland). Protoplasting efficiency was microscopically assessed every hour until the conidia fully protoplasted.

Microscopy

Bright-field and differential interference contrast microscopy were performed using a Nikon Eclipse 90i microscope (Nikon, Düsseldorf, Germany), equipped with DS-5 M camera. Aniline blue and 3,3′-diaminobenzidine (DAB) stainings were performed as described (De Neergaard, 2005).

Fluorescence microscopy was performed, using a Nikon Eclipse 90i confocal laser scanning microscope (Nikon, Düsseldorf, Germany), equipped with D-Eclipse C1-SHV camera and a pinhole diameter of 30 μm. For detection of eGFP fluorescence, an excitation wavelength of 488 nm and a 535/550 nm detection channel were employed.

Fluorescence intensity was measured using the EZ-C1 software (Nikon, Düsseldorf, Germany) as described before (Albarouki and Deising, 2013).

Image processing was done with Adobe Photoshop CS (Adobe Systems, San Jose, CA, USA).

Bioinformatic and prediction tools

In silico sequence analyses, gene deletion, gene fusion and primer construction were made using Clone Manager 9 professional (Sci-Ed., Cary, NC, USA).

Phylogenetic analyses were performed using MEGA4 software (Tamura et al., ²⁰⁰⁷). The evolutionary relationship was analysed using the Minimum Evolution method (Rzhetsky and Nei, 1992), and bootstrap consensus trees were inferred from 1000 replicates (Felsenstein, 1985).

Calculation and statistical analysis of differences between groups were done using the one-way anova test, followed by the Holm-Sidak test at a freedom degree (P < 0.001) (Sigma Plot v.11, Systat Software, Erkrath, Germany).

Accession numbers

Accession numbers of Sid1 proteins

Ajellomyces capsulatus (EEH02759.1), Arthroderma otae (XP002850162.1), Aspergillus clavatus (XP001267710.1), Aspergillus fumigatus (XP755103.1), Aspergillus nidulans (XP663427.1), Aspergillus terreus (XP001209565.1), Botryotinia fuckeliana (CCD45292.1), Claviceps purpurea (CCE29720.1), Coccidioides immitis (XP001247661.1; XP001247172.1), Coccidioides posadasii (EFW15727.1, EFW21588.1), Colletotrichum graminicola (EFQ31396.1), Colletotrichum higginsianum (CCF41271.1), Fusarium graminearum (XP385547.1), Magnaporthe oryzae (EHA47238.1), Metarhizium anisopliae (EFZ02309.1), Metarhizium acridum (EFY85959.1), Mycosphaerella graminicola (XP003855779.1), Neurospora crassa (XP960300.2), Parastagonospora nodorum (EAT90349.1), Penicillium chrysogenum (XP002558960.1), Puccinia graminis f. sp. tritici (XP003889143.1), Pyrenophora tritici-repentis (XP001938606.1; XP001942089.1; XP001932013.1), Schizosaccharomyces pombe (CAB72228.1), Sclerotinia sclerotiorum (XP001595607.1), Sordaria macrospora (XP003346089.1), Talaromyces stipitatus (XP002478751.1), Verticillium alfalfa (EEY23902.1), Verticillium dahliae (EGY14149.1).

Accession numbers of Nps6 proteins

Ajellomyces capsulatus (EEH02753.1), Ajellomyces dermatitidis (EGE78437.1), Alternaria alternata (AFN69082.1), Alternaria brassicicola (ABI51983.1), Aspergillus clavatus (XP001273566.1), Aspergillus fumigatus (EDP53310.1), Aspergillus flavus (XP002385307.1), Bipolaris carbonum (AFJ97105.1), Bipolaris oryzae (ABI51982.1), Bipolaris sorokiniana (AER36015.1), Coccidioides immitis (EAS36195.2), Coccidioides posadasii (XP003065781.1), Cochliobolus heterostrophus (AAX09988.1), Colletotrichum graminicola (EFQ32921.1), Colletotrichum gloeosporioides (ELA26594.1), Colletotrichum higginsianum (AFI23576.1), Fusarium fujikuroi (CCE73648), Fusarium graminearum (XP_383923.1), Magnaporthe oryzae (XP003714007.1), Nectria haematococca (XP003048492.1), Schizosaccharomyces pombe (CAB72227.1), Trichoderma reesei (EGR46022.1), Uncinocarpus reesii (XP002541830.1), Ustilago maydis (AAB93493.1).

Accession numbers of maize (Z . mays) defence genes

Pathogenesis-related protein 1 (PR1) (ACG29538.1), chitinase I (ACJ62153.1), GRMZM2G093346 (AFW78948.1), GRMZM2G117365 (AFW81877.1), GRMZM2G144648 (ACG24777.1) and GRMZM2G323731 (AFW80851.1).