Expressed sequence tags from the flower pathogen Claviceps purpurea

SUMMARY

The ascomycete Claviceps purpurea (ergot) is a biotrophic flower pathogen of rye and other grasses. The deleterious toxic effects of infected rye seeds on humans and grazing animals have been known since the Middle Ages. To gain further insight into the molecular basis of this disease, we generated about 10 000 expressed sequence tags (ESTs)—about 25% originating from axenic fungal culture and about 75% from tissues collected 6–20 days after infection of rye spikes. The pattern of axenic vs. in planta gene expression was compared. About 200 putative plant genes were identified within the in planta library. A high percentage of these were predicted to function in plant defence against the ergot fungus and other pathogens, for example pathogenesis‐related proteins. Potential fungal pathogenicity and virulence genes were found via comparison with the pathogen–host interaction database (PHI‐base; http://www.phi‐base.org) and with genes known to be highly expressed in the haustoria of the bean rust fungus. Comparative analysis of Claviceps and two other fungal flower pathogens (necrotrophic Fusarium graminearum and biotrophic Ustilago maydis) highlighted similarities and differences in their lifestyles, for example all three fungi have signalling components and cell wall‐degrading enzymes in their arsenal. In summary, the analysis of axenic and in planta ESTs yielded a collection of candidate genes to be evaluated for functional roles in this plant–microbe interaction.

INTRODUCTION

In the early 18th century, it was thought (incorrectly) that the ergot disease of rye and other grasses was caused by insect stings (Rapilly, 2001). It was later discovered that the biotrophic fungus Claviceps purpurea, which belongs to a large group of flower‐infecting fungi, was the actual disease incitant. Within the flower‐infecting fungi, C. purpurea belongs to a group that includes all fungal pathogens specialized in colonizing the host inflorescence and entering the host's ovary through the gynoecial pathway (Ngugi and Scherm, 2006).

The development of Claviceps on the plant is summarized in Fig. 1. In spring, wind‐dispersed ascospores are produced from ripe overwintering sclerotia at the same time as their main host—winter rye—exposes its stigmas to be pollinated. Sometimes a fungal spore instead of pollen lands on the thin‐walled stigma, which is easier to penetrate than the leaves. After germination and penetration, the fungus grows towards the bottom of the ovary and taps into the phloem sap that normally fosters the development of seeds, but now supports the growth of a white fungal mass (sphacelium) that produces conidia drenched in a sugary fluid (honeydew). Secondarily, fungal resting structures (sclerotia) replace the normal grains, whose remains are often observable on top of the ripe sclerotia.

Figure 1.

(a) Infection phases of Claviceps purpura on Secale cereale. (I) Few visible symptoms, ovary > sphacelium; 1–5 days post‐inoculation (dpi). (II) Honeydew and conidia production, ovary < sphacelium; 6–12 dpi. (III) Sclerotium development started, sphacelium > sclerotium; 13–20 dpi. (IV) Sclerotium matures, sphacelium < sclerotium; 20–50 dpi. Occurrence and relative amounts of alkaloids, lipids and carbohydrates are represented as grey bars according to Corbett et al. (1974) and our own observations. (b) Mature, ergot‐infected rye head with sclerotia.

Claviceps purpurea primarily affects the quality of harvested rye. Its sclerotia can cause severe damage to grazing animals and bread‐eating humans because of their ergot‐alkaloid content. However, ergot‐alkaloids are also produced intentionally because of their beneficial pharmaceutical activities (Krska and Crews, 2008; Schardl et al., 2006). Ergot‐alkaloids can be harvested from infected rye fields or produced in fermentors, as Claviceps can be grown in axenic culture. In axenic culture, sclerotia are never formed. Completion of the life cycle requires a living host. Therefore, C. purpurea can be called an ecologically obligate biotroph.

For a long time, we have been interested in the characterization of the axenic and pathogenic lifestyles of Claviceps, and in the identification of putative pathogenicity genes in Claviceps and putative defence genes in its host Secale cereale (Tudzynski and Scheffer, 2004). Until now, plant defence reactions towards ergot have not been reported to occur in the field.

For the efficient discovery of genes, we generated and analysed expressed sequence tags (ESTs) originating from axenic culture and from in planta tissue. ESTs are widely used to explore the transcriptome (Nagaraj et al., 2007). Currently, there are over 59 million ESTs from about 1600 species deposited in the dbEST database (http://www.ncbi.nlm.nih.gov/dbEST/). COGEME holds ESTs from 15 phytopathogenic and three saprophytic fungal species (Soanes and Talbot, 2006).

Of 9984 initial ESTs recovered from C. purpurea (one‐quarter from axenic culture and three‐quarters from in planta tissue), about 88% could be analysed. They were clustered, assembled into contigs, blasted to search for interesting genes and submitted to EMBL (Table S1, see Supporting Information). The axenic and in planta ESTs were also compared with each other. In addition, all ESTs were spotted on to nylon membranes for macroarray analyses (for example, Rolke and Tudzynski, 2008).

RESULTS AND DISCUSSION

Generation, clustering and assembly of ESTs

Three non‐normalized cDNA libraries were generated. The early in planta library, corresponding to phase I of infection, was discarded because of its low quality (see Experimental Procedures). Phase I shows little visible sign of infection and consists primarily of host tissue (Fig. 1). All clones sequenced from this library showed similarity to plant genes (data not shown). The axenic (ax) library originated from a liquid culture of C. purpurea. The in planta (ip) library corresponded to phase II and III of infection, but should contain genes expected to be expressed in phase IV, as sclerotium development initiates in phase III. Indeed, cDNAs from six genes involved in alkaloid biosynthesis (easA, easD‐G, easH1) (Schardl et al., 2006) were found in the ip library.

Sanger sequencing was performed as single‐pass reads from the 5′ end of each clone. The average length of read was 610 bases. After vector removal and sequence trimming, 8789 of the initial 9984 ESTs remained. They were clustered in order to link individual ESTs to unique genes, represented by fully, incomplete, alternatively spliced or even falsely joined transcripts. Assembly resulted in contigs, representing subsets of sequences in clusters, which overlapped without the insertion of large gaps in individual ESTs. For example, cluster 149 contained two contigs (0346 and 1224). Contig 1224 contained an intron, whereas Contig 0346 did not. In total, there were 1261 clusters and 2783 singlets. Of the 4887 contigs, 1325 contained more than one sequence.

Contigs were compared via blastx (E ≤ E‐5) against the non‐redundant protein database of the National Center for Biotechnology Information (NCBI nr): 36.5% had no match, 3.7% matched ribosomal sequences, 15.1% showed similarity to a hypothetical protein and 44.7% to a protein with known function. In addition, 29.7% had no blastx match to nr (NCBI) and no blastn match to dbEST (NCBI). These contigs may be unique to Claviceps.

Comparison of the axenic and in planta libraries

ESTs originating from two dissimilar growth conditions (ax and ip) will differ in their composition, as the fungus will react to its different surroundings, expressing some of the same and different genes. For example, this can be seen if the most abundant ax and ip ESTs are considered (Fig. 2). About one‐half of the most abundant ax ESTs are housekeeping genes: protein metabolism is represented by two ribosomal proteins, a ubiquitin fusion protein and a peptidyl–prolyl cis–trans‐isomerase, and carbohydrate metabolism by a pyruvate decarboxylase. With one exception, all highly abundant ip ESTs showed either similarity with a hypothetical protein or no similarity to a known protein. Among them, Contig 1186 may encode a protein that has recently been shown to be necessary for sexual development in Aspergillus nidulans (Han et al., 2008). It is conserved in filamentous fungi and may function as a developmental regulator. The only abundant ip EST with a putative function is Contig 0977 encoding a hydrophobin. Hydrophobins are small fungal proteins with roles in adhesion, development and pathogenesis. They were also found abundantly in EST libraries of Trichoderma harzianum (Vizcaíno et al., 2006), Laccaria bicolor and Pisolithus microcarpus (Peter et al., 2003). Interestingly, the most abundant ax and ip ESTs do not match a known protein. Yet, there are matches to EST collections of other filamentous fungi (data not shown). Some ESTs are abundant in the ax or ip library and not uncommon in the other library [for example, ESTs in Contigs 1180 (ax), 1184 and 0655 (ip)].

Figure 2.

Transcript abundance in the axenic (ax) and in planta (ip) Claviceps expressed sequence tag (EST) libraries. Light grey (dark grey) bars represent the percentage of ax (ip) sequences in the EST contigs [ax (ip) ESTs in a contig/total ax (ip) ESTs × 100]. On the left (right) side, the most abundant ax (ip) EST clones are shown. The x‐axis shows the contig number and the result of a blastx (E ≤ E‐5) search against the non‐redundant protein database of the National Center for Biotechnology Information (NCBI nr). ?, no blastx match; hyp prt, hypothetical protein; mt, mitochondrial; PDC, pyruvate decarboxylase; PPIase, peptidyl–prolyl cis–trans‐isomerase; ribosomal prt, ribosomal protein; THI, thiamine biosynthesis protein; ubiquitin fus prt, ubiquitin fusion protein; (a) blastx hit to same hyp prt, but different nucleotide sequences; (b) in Aspergillus nidulans necessary for sexual development (Han et al., 2008).

For a more detailed comparison, gene ontology (GO) terms were assigned to each EST (see Experimental procedures, Sequence analysis, GO annotation). The GO terms were counted separately for ax and ip under the aspects ‘Biological process’ and ‘Molecular function’, and significant differences were noted (see 1, 2).

Table 1.

Geneontology (GO) terms found in axenic (ax) and in planta (ip) expressed sequence tag (EST) clones—Biological process.

GO ID	Definition	ax counts/%	ip counts/%	ip counts sign. diff.
GO:0005575	Cellular component	1.00	0.79
GO:0019538	Protein metabolism	26.13	22.49	Down
GO:0006519	Amino acid and derivative metabolism	7.00	5.53	Down
GO:0006412	Protein biosynthesis (protein translation)	12.20	8.55	Down
GO:0006508	Proteolysis and peptidolysis	0.88	1.71	Up
GO:0006139	Nucleoside, nucleotide and nucleic acid metabolism	11.48	10.89
GO:0005975	Carbohydrate metabolism	3.71	5.06	Up
GO:0006629	Lipid metabolism	2.27	5.56	Up
GO:0051186	Cofactor, cofactor metabolism	1.38	1.58
GO:0009056	Catabolism	2.65	3.98
GO:0019748	Secondary metabolism	0.39	0.77
GO:0006118	Electron transport	1.22	1.13
GO:0006091	Generation of precursor metabolites and energy	3.84	4.12
GO:0007154	Cell communication	1.27	2.44	Up
GO:0016043	Cell organization and biogenesis	7.96	6.41	Down
GO:0006810	Transport	11.03	10.99
GO:0050789	Regulation of biological process	2.32	2.23
GO:0008104	Protein localization	1.71	1.98
GO:0006464	Protein modification	1.02	2.9	Up
GO:0050896	Response to stimulus	0.28	0.52
GO:0007049	Cell cycle	0.14	0.19
GO:0008283	Cell proliferation	0.06	0.05
GO:0007275	Developmental processes	0.03	0.07
GO:0006955	Immunology, immune response		0.05
GO:0046903	Secretion	0.03	0.01

ax (ip) counts/%, ax (ip) counts as a percentage of total ax (ip) GO term counts for a specific GO ID; ip counts sign. diff., number of ip counts significantly different from number of ax counts for a specific GO ID; down, ax counts > ip counts; up, ax counts < ip counts.

Table 2.

Gene ontology (GO) terms found in axenic (ax) and in planta (ip) expressed sequence tag (EST) clones—Molecular function.

GO ID	Definition	ax counts/%	ip counts/%	ip counts sign. diff.
GO:0005488	Binding	18.26	15.81
GO:0003676	Nucleic acid binding	12.70	10.17	Down
GO:0000166	Nucleotide binding	2.63	2.96
GO:0005515	Protein binding	0.82	0.50
GO:0003682	Chromatin binding	0.78	0.74
GO:0005509	Calcium ion binding	0.52	0.32
GO:0005102	Receptor binding	0.39	0.09	Down
GO:0008092	Cytoskeletal protein binding	0.04	0.23
GO:0008289	Lipid binding		0.23
	Other	0.39	0.56
GO:0003824	Catalytic activity	55.34	70.09	Up
GO:0016740	Transferase activity	16.54	17.36
GO:0016491	Oxidoreductase activity	14.51	23.23	Up
GO:0016787	Hydrolase activity	11.71	15.65	Up
GO:0016853	Isomerase activity	3.27	1.51	Down
GO:0016829	Lyase activity	2.24	3.68	Up
GO:0008233	Peptidase activity	2.20	3.81	Up
GO:0016874	Ligase activity	1.59	1.90
GO:0004386	Helicase activity	0.78	1.17
GO:0004721	Phosphoprotein phosphatase activity	0.60	0.92
GO:0004518	Nuclease activity	0.22	0.38
	Other	1.68	0.49	Down
GO:0005215	Transporter activity	7.32	7.20
GO:0005386	Carrier activity	3.32	1.55	Down
GO:0015075	Ion transporter activity	2.20	1.94
GO:0008565	Protein transporter activity	0.30	0.29
GO:0015267	Channel or pore class transporter activity		0.07
GO:0005216	Ion channel activity		0.07
	Other	1.51	3.29	Up
GO:0004871	Signal transducer activity	2.20	3.52	Up
GO:0004872	Receptor activity	0.86	1.71	Up
	Other	1.34	1.81
GO:0045182	Translation regulator activity	2.41	1.22	Down
GO:0030528	Transcription regulator activity	1.21	1.15
GO:0016209	Antioxidant activity	0.65	0.31	Down
GO:0030234	Enzyme regulator activity	0.26	0.23
GO:0003774	Motor activity	0.09
GO:0005198	Structural molecule activity	12.27	9.41	Down
GO:0003735	Structural constituent of ribosome	10.81	8.10	Down
GO:0005200	Structural constituent of cytoskeleton	0.52	0.23
	Other	0.95	1.08

Main categories are written in bold letters, subcategories not.ax (ip) counts/%, ax (ip) counts as a percentage of total ax (ip) GO term counts for a specific GO ID; ip counts sign. diff., number of ip counts significantly different from number of ax counts for a specific GO ID; down, ax counts > ip counts; up, ax counts < ip counts.

Considering biological process (Table 1), there were more counts for carbohydrate and lipid metabolism within the ip ESTs than within the ax ESTs, which is consistent with the knowledge that both types of metabolism are high in phases II and III of infection (see Fig. 1). The higher ip count for cell communication factors reflects interactions of fungal and host cells, in contrast with simpler fungal growth in culture. Surprisingly, the count for cell organization and biogenesis was higher in culture than in planta. Differences in terms related to proteins and peptides are not easily explained; lower ip counts for protein metabolism, amino acid and derivative metabolism, and protein biosynthesis corresponded to higher ip counts for proteolysis, peptidolysis and protein modification.

Table 2 summarizes the molecular function GO term counts. More catalytic activity terms were found in ip than in ax, reflecting the greater complexity of fungal growth in planta. A more detailed examination showed that, for example, isomerase activity counts were even lower for ip ESTs. In some cases, counts appeared to be contradictory: for example, ip counts for oxidoreductase activity were higher than ax counts, whereas the opposite was observed for antioxidant activity, perhaps because oxidoreductases include enzymes that produce reactive oxygen species (ROS), such as superoxide dismutases or NADPH oxidases, as well as enzymes that destroy ROS, such as catalases. This result might reflect a lower count of antioxidant activity. No evidence of an oxidative burst was observed in Claviceps‐infected rye tissue (Tudzynski and Scheffer, 2004), allowing the fungus to forego increased antioxidant activity. The count for signal transduction activity was higher in ip, consistent with the more complex milieu.

Several factors should be considered when comparing EST libraries by GO annotation. ESTs derived from the same gene could differ in length and thus appear to reflect differences between ax and ip libraries. Material for the ax library was collected from one flask, whereas material for the ip library was collected over time from many different rye spikes, thus allowing for incidental exposure to physical factors, such as temperature change, that might affect transcription. Despite these caveats, it was clear that the number of ESTs without a predicted GO term was significantly higher in the ip library than in the ax library. Similarly, there were more ESTs matching a hypothetical protein, or with no blastx match, in the ip library than in the ax library. These observations suggest that novel genes may be expressed in planta by Claviceps.

Identification of plant genes in the ip library

Plant gene identification was complicated because the rye genome has not yet been sequenced. Each ax clone was considered a priori to be of fungal origin. Any ip clone similar to an ax clone, genomic Clavicipitaceae sequences or EST sequences from Fusarium graminearum, Fusarium verticillioides or Magnaporthe grisea (fungi related to Claviceps) was considered to be putatively fungal, whereas an ip clone similar to genomic Triticeae sequences or EST sequences from the grasses S. cereale, Triticum aestivum, Hordeum vulgare or Oryza sativa was putatively considered to be of host origin (see Experimental procedures, Sequence analysis, Assignment of plant or fungal origin for ip ESTs). Of the 6545 ip clones, 3032 (46%) could be classified. Two hundred and eight ip clones (191 contigs) may have originated from the host plant (7% of the classifiable ones). All putative plant contigs were compared via blastx with the nr databank of NCBI (Table 3) to find possible functions. Forty‐two contigs had no match in the nr databank, 60 matched a hypothetical protein and seven matched a ribosomal protein. One blast hit was the S. cereale large subunit of ribulose‐1,5‐bisphosphate carboxylase/oxygenase, the plant CO₂‐binding enzyme that catalyses the first step in the Calvin cycle, thus confirming its plant origin.

Table 3.

Claviceps expressed sequence tag (EST) contigs of putative plant origin. Itemized are the results of blast searches against the non‐redundant protein database of the National Center for Biotechnology Information (NCBI nr) and two self‐assembled EST banks (flower, infected flower). Full data are only given for contigs with a meaningful hit. Contigs encoding proteins with a possible role in defence against pathogens are listed with reference.

Contig	blastx—nr (NCBI)	E‐value	seqs/ctg	blastn—flower EST bank*		blastn—infected flower EST bank*		Possible role in defence?	Reference
	best meaningful blastx hit			Best hit	E‐value	Best hit	E‐value
0115	BMY2; β‐amylase [At]	1.00E‐26	2	TC304553_Os	1.00E‐103
2495	Vacuolar ATP synthase subunit d [Nc]§	3.00E‐24	1	TC2533_Sc	0
2870	SMC1 protein [Os]	1.00E‐57	1	TC252752_Ta	0
2900	Transcription factor X1 [Tm]	5.00E‐146	1	TC149560_Hv	0
3099	Copine‐like protein [Os]	2.00E‐72	1	TC292058_Os	1.00E‐86
3359	SMC1 protein [Os]	1.00E‐94	1	TC252752_Ta	0
3751	Putative clp‐like energy‐dependent protease [Os]	9.00E‐85	1	TC292797_Os	1.00E‐141
4314	DNA‐directed RNA polymerase III largest subunit [Os]	8.00E‐67	1	TC284638_Os	9.00E‐53
4624	Plectin‐related protein‐like [Os]	3.00E‐65	1	TC287258_Os	6.00E‐74
0009	Calmodulin [Os]	4.00E‐78	2	TC234402_Ta	0	TC234402_Ta	0	Signalling	Yang and Poovaiah (2003)
0163	Triosephosphate isomerase, cytosolic [Sc]	6.00E‐121	2	TC2431_Sc	0	TC248637_TA	0
0454	Profilin‐1 [Hv]	2.00E‐61	2	TC3163_Sc	0	TC247940_Ta	0
0688	Glyceraldehyde 3‐phosphate dehydrogenase B subunit [At]	4.00E‐07	2	TC138635_Hv	1.00E‐143	TC138635_Hv	1.00E‐143
0753	Histone H2A.7 histone [Ta]	2.00E‐42	2	TC234709_Ta	0	TC235032_Ta	0
0763	Digalactosyldiacylglycerol synthase 1 [Os]	2.00E‐24	2	TC147860_Hv	1.00E‐168	TC262935_Ta	1.00E‐172
0866	Protein disulphide isomerase [Os]	3.00E‐124	2	TC139204_Hv	0	TC264921_Ta	0	Chaperone	Ray et al. (2003); Shim et al. (2004)
0905	Aspartate aminotransferase [Os]	2.00E‐68	2	TC232219_Ta	0	TC232219_Ta	0
0947	S‐Adenosylmethionine decarboxylase precursor [Ta]	0	2	TC3068_Sc	0	TC264559_Ta	0	Polyamine synthesis	Walters (2003)
1074	Adenosine kinase‐like protein [Os]	8.00E‐157	2	TC247506_Ta	0	TC247506_Ta	0
2212	Glycoprotein [Bi]	2.00E‐59	1	TC2594_Sc	0	TC247933_Ta	0
2215	Outer mitochondrial membrane protein porin [Ta]	1.00E‐106	1	TC146872_Hv	0	TC250606_Ta	0
2272	Polyubiquitin [Ss]	7.00E‐45	1	TC234281_Ta	0	TC234209_Ta	0	Protein degradation	Dreher and Callis (2007)
2279	Histone H4 [Ho]	1.00E‐38	1	TC233179_Ta	0	TC250187_Ta	1.00E‐151
2365	MRP‐like ABC transporter [Os]	1.00E‐129	1	TC354124_Os	0	TC237754_Ta	0
2369	Boron transporter [Os]	1.00E‐22	1	TC252478_Ta	1.00E‐173	TC252478_Ta	1.00E‐173
2392	DEAD/DEAH box helicase, putative [Os]	5.00E‐123	1	TC307499_Os	0	TC271590_Ta	8.00E‐52
2415	Glutamine synthetase isoform GS1b [Ta]	4.00E‐136	1	TC139434_Hv	0	TC264890_Ta	0
2458	Polyubiquitin [Ss]	9.00E‐102	1	TC2362_Sc	0	TC234234_Ta	0	Protein degradation	Dreher and Callis (2007)
2560	LON1 protease [Ta]	2.00E‐129	1	TC131081_Hv	0	TC235913_Ta	0
2573	RuBisCO large subunit‐binding protein subunit beta [Sc]	8.00E‐120	1	TC139132_Hv	0	TC249610_Ta	0
2683	Glucose‐6‐phosphate dehydrogenase [Ta]	6.00E‐64	1	TC139350_Hv	0	TC251298_Ta	0
2687	Metallothionein [Ta]	3.00E‐07	1	TC246696_Ta	1.00E‐127	TC246696_Ta	1.00E‐127	ROS scavenger	Zhou et al. (2006)
2699	16.9‐kDa class I heat shock protein [Ta]	1.00E‐70	1	TC249526_Ta	0	TC140216_Hv	0	General stress response	Vierling (1991); Wang et al. (2004)
2768	Elongation factor 2 [Os]	3.00E‐116	1	TC3094_Sc	0	TC246958_Ta	0
2771	Calmodulin TaCaM1‐1 [Os]	5.00E‐77	1	TC330697_Os	0	TC237135_Ta	0	Signalling	Yang and Poovaiah (2003)
2802	Superoxide dismutase 1, putative [Os]	9.00E‐75	1	TC2466_Sc	0	TC262714_Ta	0	ROS scavenger	Mittler et al. (2004)
2837	α‐Glucosidase II [Os]	2.00E‐126	1	TC317731_Os	0	TC133712_Hv	0
2901	AAA‐superfamily of ATPases [Os]	1.00E‐54	1	TC301332_Os	7.00E‐55	TC271109_Ta	9.00E‐86
3014	Ubiquitin‐like protein SMT3 [Os]	7.00E‐34	1	TC2462_Sc	0	TC250842_Ta	0	Protein degradation	Dreher and Callis (2007)
3101	26S proteasome subunit 4‐like protein[Os]	6.00E‐112	1	TC146596_Hv	0	TC146596_Hv	0	Protein degradation	Dreher and Callis (2007)
3132	Remorin [Os]	1.00E‐22	1	TC329748_Os	5.00E‐68	TC252212_Ta	3.00E‐67	Response to (a)biotic stimuli	Raffaele et al. (2007)
3172	APETALA3‐like protein [Hv]	2.00E‐34	1	TC2593_Sc	1.00E‐137	TC238728_Ta	1.00E‐119
3228	H‐ATPase [Os]	3.00E‐63	1	TC247104_Ta	0	TC248557_Ta	0
3239	Cytosolic glutathione peroxidase [Tm]	2.00E‐92	1	TC131656_Hv	0	TC246747_Ta	0	ROS scavenger	Mittler et al. (2004)
3288	Cysteine protease [Ta]	2.00E‐11	1	TC232622_Ta	1.00E‐88	TC232303_Ta	7.00E‐91
3374	Chitinase 1 precursor, type III [Os]	3.00E‐111	1	TC147411_Hv	0	TC266129_Ta	0	PR protein	van Loon et al. (2006)
3378	Ascorbate peroxidase [Hv]	2.00E‐120	1	TC233791_Ta	0	TC232978_Ta	0	ROS scavenger	Mittler et al. (2004)
3477	Thioredoxin h isoform 2; HvTrxh2 [Hv]	2.00E‐52	1	TC3164_Sc	0	TC267391_Ta	0	ROS scavenger	Mittler et al. (2004)
3502	Glucose‐6‐phosphate dehydrogenase [Ta]	6.00E‐82	1	TC139350_Hv	0	TC251298_Ta	0
3535	26S protease regulatory subunit 7 [Os]	4.00E‐134	1	TC2796_Sc	0	TC246888_Ta	0	Protein degradation	Dreher and Callis (2007)
3545	Polyubiquitin 2 [Da]	3.00E‐66	1	TC92012_Sb	0	TC232408_Ta	0	Protein degradation	Dreher and Callis (2007)
3559	Myo‐inositol 1‐phosphate synthase [Zm]	2.00E‐69	1	TC2647_Sc	0	TC246900_Ta	0
3706	Putative heat shock 70 KD protein, mitochondrial [Os]	1.00E‐15	1	TC139412_Hv	1.00E‐79	TC264160_Ta	1.00E‐113	General stress response	Vierling (1991); Wang et al. (2004)
3714	Lipid transfer protein‐like protein 2 precursor [Sc]	4.00E‐35	1	TC265011_Ta	5.00E‐67	TC234771_Ta	1.00E‐75	PR protein	van Loon et al. (2006)
3791	Triticain β[Ta]	2.00E‐38	1	TC147127_Hv	1.00E‐150	TC233092_Ta	1.00E‐156	Programmed cell death and others	Goodwin et al. (2004); Matarasso et al. (2005)
3848	Stearoyl‐ACP desaturase [Os]	1.00E‐66	1	TC139428_Hv	0	TC235225_Ta	0	Signalling	Kachroo et al. (2007)
3858	Ubiquitin [Os]	8.00E‐66	1	TC2357_Sc	0	TC249401_Ta	0	Protein degradation	Dreher and Callis (2007)
3869	Hypersensitive‐induced reaction protein 3 [Hv]	8.00E‐33	1	TC247941_Ta	1.00E‐170	TC247941_Ta	1.00E‐171	Hypersensitive response	Rostoks et al. (2003)
3997	Proteinase inhibitor‐related protein [Ta]	3.00E‐36	1	TC234467_Ta	1.00E‐158	TC234468_Ta	1.00E‐178	PR protein	van Loon et al. (2006)
4134	2‐Oxoglutarate/malate translocator [Pm]	1.00E‐56	1	TC3237_Sc	0	TC264258_Ta	0
4137	Ribonucleotide reductase [Os]	6.00E‐06	1	TC3660_Sc	8.00E‐63	TC265616_Ta	1.00E‐124
4163	Glutathione‐S‐transferase, class Tau [Os]†	5.00E‐75	1	TC326296_Os	4.00E‐53	TC265501_Ta	0	ROS scavenger	Dixon et al. (2002)
4204	Histone H2B.2 [Os]	1.00E‐39	1	TC3095_Sc	0	TC264367_Ta	0
4217	Ubiquitin‐conjugating enzyme [Ta]	1.00E‐18	1	TC2608_Sc	0	TC232178_Ta	1.00E‐135	Protein degradation	Dreher and Callis (2007)
4326	S‐Adenosyl‐l‐homocysteine hydrolase [Hv]	9.00E‐151	1	TC139066_Hv	0	TC264652_Ta	0
4345	Small Ras‐related GTP‐binding protein [Ta]	1.00E‐113	1	TC3107_Sc	0	TC234441_Ta	0
4356	Elongation factor 1β[Hv]	2.00E‐84	1	TC131466_Hv	0	TC131466_Hv	0
4388	Cap‐binding protein [Os]	3.00E‐36	1	TC251778_Ta	0	TC251778_Ta	0
4390	Class III alcohol dehydrogenase enzyme [Os]	3.00E‐94	1	TC263255_Ta	0	TC262835_Ta	0
4528	Small nuclear ribonucleoprotein homologue [Os]	2.00E‐39	1	TC341008_Os	1.00E‐103	TC249742_Ta	0
4529	Cytochrome P450 like_TBP [Nt]	2.00E‐11	1	TC262719_Ta	3.00E‐80	TC262719_Ta	1.00E‐80
4568	Ubiquitin carrier protein [Os]	5.00E‐82	1	TC3331_Sc	0	TC247094_Ta	0
4574	Endoplasmin homologue precursor [Hv]	3.00E‐29	1	TC264521_Ta	0	TC264521_Ta	0
4673	Phospho‐2‐dehydro‐3‐deoxyheptonate aldolase [Os]	2.00E‐11	1	TC234760_Ta	0	TC234760_Ta	0
4680	Phytepsin precursor [Hv]	2.00E‐139	1	TC3124_Sc	0	TC232293_Ta	0	Diverse	Schaller (2004)
4721	Alanine aminotransferase 2 [Hv]	1.00E‐114	1	TC146731_Hv	0	TC264334_Ta	0
4762	1,4‐α‐d‐glucan 6‐α‐d‐(1,4‐α‐d‐glucanotransferase [Ta]	3.00E‐90	1	TC139565_Hv	0	TC236593_Ta	0
2956	Prolyl 4‐hydroxylase α subunit homologues [Os]	2.00E‐80	1			TC254333_Ta	0	Cell wall fortification	Deepak et al. (2007)
3042	FAD‐linked oxidase family protein [At]‡	1.00E‐92	1			TC140734_Hv	1.00E‐105	ROS producer	Taler et al. (2004)
2431	Ubiquitin‐associated protein [At]	9.00E‐26	1					Protein degradation	Dreher and Callis (2007)
2913	Single‐strand DNA repair‐like protein [Tm]	2.00E‐19	1
3280	UPL4 (ubiquitin‐protein ligase 4) [At]	7.00E‐25	1					Protein degradation	Dreher and Callis (2007)
3537	UDP‐glucuronosyl/UDP‐glucosyltransferase family protein [Os]	5.00E‐12	1					Stress response	Gachon et al. (2005)
3579	Elongation factor Tu, mitochondrial precursor [Ate]¶	4.00E‐87	1
3854	Calmodulin‐like [Os]	2.00E‐35	1					Signalling	Yang and Poovaiah (2003)
3944	Plastid division regulator MinE [Hv]	4.00E‐20	1
4168	Lipolytic enzyme, G‐D‐S‐L family protein [Os]	2.00E‐30	1
4195	Thaumatin‐like protein [Ta]	3.00E‐69	1					PR protein	van Loon et al. (2006)
4539	Putative RNA apurinic site specific lyase [Os]	6.00E‐21	1

In addition: 7 contigs—ribosomal proteins; 57 contigs—hypothetical proteins (four hits only with infected flower EST bank); 38 contigs—no match (eight hits only with infected flower EST bank).

Organism abbreviations: At, Arabidopsis thaliana; Ate, Aspergillus terreus; Bi, Bromus inermis; Da, Deschampsia antarctica; Ho, Hyacinthus orientalis; Hv, Hordeum vulgare; Nc, Neurospora crassa; Nt, Nicotiana tabacum; Os, Oryza sativa; Pm, Panicum miliaceum; Sb, Sorghum bicolor; Sc, Secale cereale; Sp, Sporobolus stapfianus; Ta, Triticum aestivum; Tm, Triticum monococcum; Zm, Zea mays.

^* Self‐assembled EST banks (see Experimental procedures) (‘TC identifier’_‘abbreviated organism’).

^† Glutathione peroxidase activity/often induced by infection.

^‡ GO:0008891 glycolate oxidase activity.

^§ Probably joined cDNA.

^¶ Match for typing as ‘putatively plant’: WHEAT‐146134, 1.00E‐147.

Contigs with putative functions in plant defence were noted, with the caveat that in no case was a defined role documented in this system. Some contigs contained components of the ubiquitin/26S proteosome protein degradation system; those found are not known to be required for resistance, but plants generally increase their ubiquitylation capacity after microbial attack (Dreher and Callis, 2007). Heat shock proteins (HSP20 of class I and HSP70) are considered to be part of a general stress response, whereas remorin (a plant‐specific plasma membrane protein) responds to abiotic and biotic stimuli (Raffaele et al., 2007). In addition, UDP‐glucuronosyl transferases, which have a function in anthocyanin biosynthesis, have been suggested to play a role in stress tolerance (Gachon et al., 2005). A UDP‐glucuronosyl transferase was found to be a basal defence transcript in the interaction of wheat with Puccinia striiformis f. sp. tritici (Coram et al., 2008). In addition, disulfide isomerase was found; it belongs to a protein family that responds in plants and animals to stress conditions associated with enhanced chaperone activity. The enzyme has been investigated in two interactions of a grass and a hemibiotrophic fungus (wheat—Mycosphaerella graminicola by Ray et al., 2003; rice—M. grisea by Shim et al., 2004). General response elements involving signalling were found: two calmodulins and a calmodulin‐like protein. Calmodulins may play a role in defence‐related calcium signalling (Yang and Poovaiah, 2003). In addition, a stearyl‐ACP desaturase was found, an archetypical member of a family of soluble fatty acid desaturases. Stearyl‐ACP desaturase mutants in Arabidopsis thaliana have been shown to be impaired in the jasmonic acid defence signalling pathway (2007, 2001). Also identified was S‐adenosylmethionine decarboxylase, a key polyamine biosynthetic enzyme. Polyamines could be involved in triggering the hypersensitive response (HR) (Walters, 2003).

A connection for HR and programmed cell death (PCD) was suggested for two other contigs. Contig 3869 shows similarity to hypersensitive‐induced reaction protein 3 from barley, which is increased 35‐fold in mutants with a spontaneous HR phenotype (Rostoks et al., 2003). Contig 3791 may encode a triticain‐like cysteine protease. Cysteine proteases have been connected with some features of plant defence, including PCD (Goodwin et al., 2004; Matarasso et al., 2005). Another putative protease shows similarity to phytepsin, an aspartic proteinase implicated in defence against microbial pathogens and in pollen–pistil interaction (Schaller, 2004). A more direct effect may involve the action of a prolyl‐4‐hydrolase, which hydroxylates proline‐rich structural glycoproteins of the cell wall, contributing to the physical barrier against incoming pathogens. An increase in such glycoproteins was found in pearl millet infected with Sclerospora and in wheat or maize infected with Fusarium (Deepak et al., 2007). ROS typify the plant–pathogen interaction (Mittler et al., 2004; Torres et al., 2006). Several putative ROS scavengers were found: superoxide dismutase, glutathione peroxidase, ascorbate peroxidase, thioredoxin and metallothionin. A class tau glutathione‐S‐transferase was also found, which may be a ROS scavenger, as some theta, pi and tau glutathione‐S‐transferases have been shown to have glutathione peroxidase activity (Dixon et al., 2002). Only one possible ROS producer was found, a flavin adenine dinucleotide (FAD)‐linked oxidase family protein, which has d‐lactate dehydrogenase as well as glycolate oxidase activity according to GO classification. Glycolate oxidase, a key enzyme of photorespiration and an H₂O₂ producer, has been associated with the protection of melon against downy mildew (Taler et al., 2004). Something similar was observed by Bohman et al. (2002) for Leptosphaeria infecting a chimeric Brassica/Arabidopsis plant. The finding of ROS‐related EST contigs is interesting, because rye ovaries exhibit an oxidative burst after infection with a CREB‐like transcription factor knockout mutant (Nathues et al., 2004) and a small GTPase knockout mutant of C. purpurea (Scheffer et al., 2005). Pathogenesis‐related (PR) proteins (van Loon et al., 2006) are common responses of plants to attack. Within the ip clones, there appear to be a PR‐5 (a thaumatin‐like protein), a PR‐6 (a protease inhibitor of the Bowman–Birk type), a PR‐8 (a chitinase type III) and a PR‐14 (a non‐specific lipid transfer protein).

Interestingly, pathogenicity‐related EST sequences were also found in the transcriptome analysis of the close relative of C. purpurea, Neotyphodium (Felitti et al., 2006; http://hornbill.cspp.latrobe.edu.au/cgi‐binpub/endophyte/index.pl). Some ESTs are found only in an interaction of Neotyphodium with a grass. Some most probably originate from the plant partner. As in the Claviceps—rye interaction, a Bowman–Birk‐type wound‐induced proteinase inhibitor and non‐specific lipid transfer proteins were found. In addition, there are hits to germin, known to be involved in the cereal defence response against invading fungi (Lane, 2002), and the PR protein AOPR1 from Asparagus officinalis. Of course, Neotyphodium (the anamorph of Epichloë) is considered by some to be a harmless endophyte, despite the fact that its alkaloids poison grazing livestock.

Because Claviceps is a flower pathogen, it is intriguing that many PR proteins were found to be constitutively present in floral tissues. In general, one finds the expression of defence‐related genes in flowers (Laitinen et al., 2005; Liljeroth et al., 2005; Tung et al., 2005). Flowers would otherwise be extremely good places for a pathogen to attack, as pistils have no cuticle or only a disrupted one. Defence‐related contigs were relatively abundant (19% of all putative plant contigs, 40% of all those with an assigned possible function). Host tissue in this disease is of flower origin, distinct because it must encourage pollen tube penetration but repel pathogens. To address this, all putative plant contigs were compared via blastn with two self‐assembled EST banks of grasses. The ‘flower’ bank comprised ESTs from pollen, anthers, pollinated pistils, pollinated spikes, spikelets and panicles of millet, rye, barley, rice, wheat and maize. The ‘infected flower’ bank contained ESTs from spikes and heads of barley and wheat infected with the necrotroph F. graminearum (Experimental procedures, Sequence analysis). The largest group in Table 3 showed matches to ESTs in both banks, seeming to confirm the preparedness of flowers for pathogens. Most (79%) of all putative plant contigs showed significant similarity to the grass flower EST bank, which is consistent with our observations of infected rye spikes. Some contigs in Table 3 had no hit to either flower EST bank, but several of these contigs had a possible defence role (for example, Contigs 3537 and 4195). There were just two contigs with a hit only to the infected flower EST bank (Contigs 2956 and 3042). This interesting group can be enlarged by four contigs matching hypothetical proteins and eight contigs with no match at all. The last group is small, showing hits only to the flower EST bank and containing no contig with a possible role in defence.

Ergot is a replacement disease and our EST library does not contain clones from the first 5 days (phase I, Fig. 1), when the ovaries are barely invaded by the fungus. Therefore, few plant genes were expected within the library. Any plant genes found are interesting as the harvested part (infected ovary) included the rachilla, the site at which fungus and plant form a stable plant–pathogen interface. The defence genes found may be there because flowers by themselves are always on guard or because rye—contrary to current doctrine—recognizes the invading fungus and the fungus is able to suppress and balance the plant defences. In the interaction of Ustilago and maize, it was thought that classical defence pathways were not activated, but Doehlemann et al. (2008) have shown recently that, after the infection of seedlings, U. maydis is recognized early and triggers defence responses.

Identification of pathogenicity genes

To find potential pathogenicity and virulence genes in our EST collection, we compared the proteins of the pathogen–host interaction database (PHI‐base; (Baldwin et al., 2006; Winnenburg et al., 2008) with our Claviceps ESTs. PHI‐base contains curated information on pathogenicity, virulence and effector genes of mostly fungal pathogens. We proceeded on the assumption that the proof of pathogenicity/virulence function of a gene in one fungus also suggests its pathogenicity/virulence function in other fungi (Baldwin et al., 2006).

In accordance with this assumption, the search of PHI‐base proteins yielded several already characterized C. purpurea molecules. Eight EST contigs were identical with already cloned genes (Contigs 0213, 0923, 2567, 2751, 1398, 0448, 1323 and 1513). Six of these were either pathogenicity or virulence factors: mitogen‐activated protein (MAP) kinases cpmk1 and cpmk2, PAK kinase cla4, histidine kinase cphk2, polygalacturonase 2 and bZip transcription factor cptf1. Three EST contigs were related to already cloned genes (Contigs 2301, 0148 and 3103), two of which were pathogenicity factors (serine/threonine kinase cot1 and Rho GTPase cdc42).

The assumption is also valid when comparing fungal plant and animal pathogens. Sexton and Howlett (2006) found—based on the December 2006 PHI‐base content and with a focus on ascomycetes—that there are parallels in the infection mechanisms used by fungal plant and animal pathogens. Many of those highlighted by Sexton and Howlett (2006) were also found as potential virulence or pathogenicity genes in the pool of Claviceps ax and ip ESTs (Table 4). During germination and during in planta growth, signalling components, such as MAP kinase, guanosine triphosphatases (GTPases) and cyclic adenosine monophosphate (cAMP)‐dependent kinases, are active (Table 5, ‘Signalling’ section). Isocitrate lyase is thought to be involved in providing the pathogen with nutrition (Table 4, ‘Primary metabolism’ section). Some hydrophobins are needed for infection (Table 4, ‘Hydrophobins and related’ section), although that found in C. purpurea is a pentahydrophobin, already proven not to affect pathogenicity under the conditions investigated. The biosynthesis of melanin—a fungal cell wall component—is often important in the penetration stage (Table 4, ‘Melanin synthesis’ section). In addition, other fungal cell components, for example chitin or β‐glucan, are important (Table 4, ‘Fungal cell wall synthesis/modification’ section). Hydrolytic enzymes are at least partly essential during infection (Table 4, ‘Plant cell wall‐degrading enzymes’ and ‘Proteases’ sections). ROS‐scavenging enzymes may restrict host‐produced ROS damage or tune ROS‐dependent signal transduction (Table 4, ‘ROS and related’ section). Our hypothesis is that Claviceps produces ROS, but keeps net ROS production at a low level (Nathues et al., 2004).

Table 4.

Claviceps expressed sequence tag (EST) contigs with significant homology to pathogen–host interaction database (PHI‐base) entries. The number and origin of the EST clones assembled in each contig are indicated: ax (ip) seqs, axenic (in planta) sequences.

Contig	ax seqs	ip seqs	Putative function	Best PHI‐base match						Other PHI‐base matches
				PHI	Species*	Host†	Phenotype‡	HSP	E‐value	# PHI	Hosts§	Phenotypes¶
			Signalling
3832		1	G α subunit	1013	F. graminearum	gr	vir	157	7.00E‐40	7	2 gr/3 pl/2 an	vir/path/unaff
3221		1	MAPKK kinase	1016	F. graminearum	gr/pl	path	408	E‐115	5	3 gr/ 2an	vir/path
0213		2	MAP kinase (cpmk1)	245††	C. purpurea	gr	path	274	E‐145	15	7 gr/6 pl/2 an	path/vir/unaff
0923	1	1	MAP kinase (cpmk2)	246††	C. purpurea	gr	vir	765	0	5	2 gr/2 pl/2 an	vir/path
2567		1	PAK kinase	cla4**, ††	C. purpurea	gr	path	451	E‐127	4	2 gr/3 an	vir/path/unaff/lethal
2751		1	Histidine kinase	cphk2**, ††	C. purpurea	gr	vir	114	5.00E‐26
2516		1	HOG1‐type MAP kinase	153	M. grisea	gr	unaff	375	E‐105	4	1 gr/2 pl/2 an	vir/unaff
3498		1	PKA, catalytic subunit	36	M. grisea	gr	path	572	E‐164	6	1 gr/2 pl/3 an	path/vir
0806	5	2	G β subunit	334	C. heterostrophus	gr	path	90	2.00E‐19	2	2 pl	vir
2301		1	Serine/threonine kinase	158	U. maydis	gr	vir	218	1.00E‐57	1‡‡	1 gr	path
0738		2	Guanyl nucleotide exchange factor	319	U. maydis	gr	vir	117	6.00E‐27
1835	1		Adenylate cyclase	332	C. lagenarium	pl	path	198	3.00E‐51	2	1 gr/1 pl	vir
3604		1	PKA, regulatory subunit	231	C. lagenarium	pl	path	134	1.00E‐32	1	1 gr	unaff
0276		3	Rab/GTPase	339	C. lindemuthianum	pl	vir	286	5.00E‐98
0148		3	Rho GTPase	270	C. albicans	an	path	300	4.00E‐83	4‡‡	2 gr	path/vir
0124		2	CDC2‐related/cyclin‐dependent protein kinase	172	C. albicans	an	path	169	7.00E‐43
1564	1		Type 2A‐related protein phosphatase	378	C. albicans	an	vir	181	5.00E‐47
3000		1	Phospholipase D	373	C. albicans	an	vir	228	2.00E‐60
0270	1	2	Ras GTPase	182	C. neoformans	an	vir/path	293	5.00E‐81	3	2 gr/1 an	vir/path
1031	13	7	Cyclophilin	213	C. neoformans	an	vir	266	4.00E‐73	2	1 gr/1 pl	vir
1471	1		Calcineurin A catalytic subunit	89	C. neoformans	an	path	264	1.00E‐71	1	1 an	vir
3516		1	Calcineurin B regulatory subunit	474	C. neoformans	an	path	120	5.00E‐31	1	1 an	vir
2970		1	Rho guanyl‐nucleotide exchange factor	675	C. neoformans	an	hypvir	276	7.00E‐75
1118		2	Rheb GTPase	317	A. fumigatus	an	vir	228	2.00E‐61
2239		1	Fatty acid oxygenase	494	A. fumigatus	an	hypvir	117	3.00E‐27
3582		1	Inositolphosphorylceramide synthase	218	C. neoformans	an	vir	229	2.00E‐61
			Transporter
2879		1	ABC transporter	391/1018	M. grisea	gr	unaff	292	1.00E‐79	3	2 gr/ 2pl	vir/unaff
0188		2	Monocarboxylate permease	812	M. grisea	gr	vir	90	2.00E‐19
4063		1	MFS transporter	737	C. nicotianae	pl	vir	151	7.00E‐38	1	an	vir
2267		1	Transporter	544	B. cinerea	pl	unaff	244	7.00E‐66	2	1 gr/1 pl	vir/lethal
1245	1	11	Mitochondrial carrier protein	254	F. oxysporum	pl	vir	545	E‐156	1	1 gr	vir
3299		1	Sugar transporter	689	C. neoformans	an	unaff	115	6.00E‐27	1	1 pl	unaff
2452		1	Voltage‐gated chloride channel	286	C. neoformans	an	vir	123	4.00E‐29
1365	1		Ferrichrome‐type siderophore transporter	513	C. albicans	an	vir/unaff	214	1.00E‐56
0997	1	2	Iron permease	162/485	C. albicans	an	path	176	3.00E‐45
1015	1	1	Vacuolar H(+)‐ATPase	435	C. albicans	an	path	139	5.00E‐35
			β‐Glucosidases
3076		1	β‐Glucosidase	748	U. maydis	gr	unaff	224	2.00E‐59	2	1 gr/2 pl	unaff/path
0352	1	2	β‐Glucosidase	816	M. grisea	gr	vir	246	3.00E‐66
			Fungal cell wall synthesis/modification
1966	1		Chitin synthase	389	U. maydis	gr	vir	206	4.00E‐54	3	1 pl/3 an	vir/path
1551	1		1,3‐β‐Glucanosyltransferase	522	F. oxysporum	pl	vir	379	E‐106	1	1 an	vir
4679		1	Chitin synthase	236	W. dermatitidis	an	vir	143	5.00E‐35	4	2 pl/2 an	vir/unaff
0893		2	1,3‐β‐Glucanosyltransferase	434	A. fumigatus	an	vir	168	4.00E‐69
0560	2	1	Cell‐surface glycosidase	347	C. albicans	an	vir	85	3.00E‐24
4592		1	Mannosyl transferase	392	C. albicans	an	vir	241	8.00E‐65
1590	1		Protein mannosyltransferase	454	C. albicans	an	vir	114	2.00E‐26
1111	1	1	Mannose‐6‐phosphate isomerase	220	C. neoformans	an	vir	299	2.00E‐82
2424		1	Calcium/manganese P‐type ATPase	440	C. albicans	an	vir	194	2.00E‐50
0283	1	1	Chitinase	144	T. virens	fungus	vir	182	6.00E‐49
			Plant cell wall‐degrading enzymes
1398	1		Polygalacturonase	437††	C. purpurea	gr	path	491	E‐140	4	1 gr/3 pl	vir/unaff
1166		2	Cellulase	566	C. carbonum	gr	unaff	115	1.00E‐33
0403		2	Pectin methyl esterase	278	B. cinerea	pl	vir/unaff	193	1.00E‐50
			ROS and related
0448	4	4	Superoxide dismutase [Cu–Zn]	769††	C. purpurea	gr	unaff	291	1.00E‐81	4	1 pl/3 an	vir/path
3103		1	Catalase	542	B. cinerea	pl	unaff	202	7.00E‐53	2/cat1**, ‡‡	2 gr/1an	vir/unaff
0200	1	5	Superoxide dismutase [Mn]	401	C. albicans	an	unaff	236	8.00E‐64	2	2 an	path
1874	1		Glutathione reductase	692	C. neoformans	an	path	164	8.00E‐43
1279	2	4	GSNO reductase	668	C. neoformans	an	vir/unaff	355	E‐148
			Proteases
0561	2	1	Signal peptidase	248	C. graminicola	gr	path	317	5.00E‐88
0180	1	9	Aspartyl proteinase	697	L. maculans	pl	unaff	568	E‐163	2	2 an	vir
3092		1	Metalloprotease 1 precursor	479	C. posadasii	an	vir	194	6.00E‐51
			PKS and NRPS (secondary metabolism)
2248		1	Polyketide synthase	726	F. graminearum	gr	unaff	225	6.00E‐86	4	2 gr/1 pl/1 an	vir/effec
2989		1	Polyketide synthase	714	F. graminearum	gr	unaff	170	7.00E‐43
1068		2	Type I polyketide synthase	722	F. graminearum	gr	unaff	132	1.00E‐31
2483		1	Polyketide synthase	255	G. moniliformis	gr	unaff	169	2.00E‐42
2957		1	Polyketide synthase	40	C. lagenarium	pl	vir	131	3.00E‐31	2	2 an	vir
2781		1	Non‐ribosomal peptide synthetase	745	A. brassicicola	pl	vir	100	5.00E‐21	3	3 gr	vir
			Sterol and isoprenoid biosynthesis
0840		2	Hydroxymethyl‐glutaryl CoA reductase	1006	F. graminearum	gr	vir	476	E‐135
			Melanin synthesis
3472		1	Carnitine acetyl transferase	120/594	M. grisea	gr	path	263	7.00E‐74
1176		2	Scytalone dehydratase	58	C. lagenarium	pl	vir	207	2.00E‐55
			Primary metabolism
1678	1		Trehalose‐6‐phosphate synthase	322	M. grisea	gr	path	237	1.00E‐63	2	1 pl/1an	vir/unaff
0098		2	Isocitrate lyase	305	M. grisea	gr	vir	172	3.00E‐44	2	1 pl/1 an	vir
0472	1	1	S‐Adenosylmethionine synthetase	877	M. grisea	gr	vir	367	E‐103	1	1 an	path
3344		1	Imidazole glycerol phosphate dehydratase	121	M. grisea	gr	path	323	4.00E‐90
2470		1	β‐Isopropyl‐malate dehydrogenase	415	S. nodorum	gr	path	385	E‐108	1	1 an	vir
2523		1	Glyoxylase I	414	S. nodorum	gr	unaff	333	1.00E‐95
1488	1		δ‐Aminolevulinic acid synthase	595	S. nodorum	gr	vir	259	4.00E‐70
0837	5	1	Methionine synthase	442	F. graminearum	gr	vir	851	0
2730		1	Homoserine O‐acetyltrasferase	355	F. graminearum	gr	vir	281	4.00E‐77
0049	1	1	Cystathionine β‐lyases	443	F. graminearum	gr	vir	567	E‐163
0942		2	Fatty acid synthase β subunit	97	C. carbonum	gr	vir	253	E‐121
1368	1		Hexokinase	869	B. cinerea	pl	vir	486	E‐139
0857		2	Argininosuccinate lyase	200	F. oxysporum	pl	vir	311	2.00E‐91
4270		1	Alcohol oxidase	199	C. fulvum	pl	vir	153	3.00E‐38
3094		1	3‐Ketoacyl‐CoA thiolase	598	L. maculans	pl	vir	97	2.00E‐21
0853	1	1	Phosphoenolpyruvate carboxykinase	424	C. neoformans	an	vir	142	6.00E‐35
0432	2	2	Sulphate adenyltransferase	265	C. neoformans	an	path	431	E‐122
2916		1	Acetolactate synthase	358	C. neoformans	an	path	192	5.00E‐50
4363		1	Methylcitrate synthase	447	A. fumigatus	an	vir	387	E‐109
3678		1	Fatty acid synthase alpha subunit	96	C. albicans	an	path	162	2.00E‐48
0666		3	Orotidine‐5′‐phosphate decarboxylase	506	S. cerevisiae	an	vir	92	2.00E‐20
			Structural molecules
0251	5	7	Woronin body major protein	356	M. grisea	gr	vir	260	4.00E‐71
0266		5	Cell wall protein	175	C. albicans	an	vir	100	3.00E‐22
3575		1	Septin	281	C. albicans	an	vir	263	7.00E‐73
0366		2	Septin	282	C. albicans	an	vir	141	6.00E‐35
			Hydrophobins and related
1323		5	Hydrophobin	291††	C. purpurea	gr	unaff	575	E‐165	1	1 gr	vir
2720		1	snodprot1 homologue	860	M. grisea	gr	vir	430	E‐122
0273		6	Cerato‐platanin	695	L. maculans	pl	unaff	127	3.00E‐31
			Transcription factors
1200		6	Transcription factor	776/802	M. grisea	gr	vir	131	9.00E‐32
1513	1		bZip transcription factor	344††	C. purpurea	gr	vir	365	E‐102
1606	1		bZIP transcription factor	444	F. graminearum	gr	vir	120	2.00E‐30
0516	1	1	Cross‐pathway control protein	467	C. parasitica	pl	vir	150	9.00E‐38	1	1 an	vir
4589		1	Ste12‐like transcription factor	294	C. lagenarium	pl	path	113	4.00E‐26	2	1 gr/1 pl	path
2296		1	Regulator of filamentous growth/virulence	475	C. albicans	an	path	90	3.00E‐19
			Cell cycle
0968	1	1	G1 cyclin	524	U. maydis	gr	path/unaff	204	9.00E‐54
3793		1	Cell division control protein	773/791	M. grisea	gr	vir	179	9.00E‐46
2107	1		Cell cycle regulatory protein	346	U. maydis	gr	vir	213	2.00E‐56
			DNA/RNA/chromatin structure
2521		1	Histone deacetylase	260	U. maydis	gr	unaff	234	1.00E‐62	1	1 gr	vir
1626	1		Related to DNA repair protein Rad	890	M. grisea	gr	vir	245	6.00E‐66
2592		1	Topoisomerase I	80	C. albicans	an	vir	182	6.00E‐47
			Heat shock proteins
0198	6	9	Heat shock protein HSP70	682	C. neoformans	an	vir	645	0
0755		2	Heat shock protein HSP90	463	S. cerevisiae	an	hypvir	351	9.00E‐98
			Assorted
1276		4	Ornithine decarboxylase	177	S. nodorum	gr	vir	265	E‐131
1167	8	12	Zn‐binding alcohol dehydrogenase	881	M. grisea	gr	vir	150	8.00E‐38
1944	1		Reverse transcriptase	876	M. grisea	gr	vir	192	1.00E‐49
0573		3	Unknown (carbonic anhydrase domain)	888	M. grisea	gr	vir	369	E‐103
0374		3	Unknown (LisH motif )	806	M. grisea	gr	vir	238	2.00E‐64
2903		1	Unknown (six‐hairpin glycosidase‐like)	785	M. grisea	gr	vir	147	2.00E‐36
0399	1	2	Short‐chain dehydrogenase/reductase	784	M. grisea	gr	vir	126	2.00E‐30
4415		1	Ribonuclease	811	M. grisea	gr	vir	124	7.00E‐30
0723	3		Mitochondrial glycoprotein of p32 family	367	U. maydis	gr	vir	135	3.00E‐33
3155		1	Extracellular lipase	432	F. graminearum	gr	vir	60	1.00E‐18
0821		10	Perilipin‐like prt	27	C. gloeosporioides	pl	path	168	2.00E‐43
0326	2	1	FKBP‐type peptidyl‐prolyl cis‐trans isomerase	548	B. cinerea	pl	vir	162	5.00E‐42
3174		1	Cytochrome P450 monooxygenase	438	B. cinerea	pl	vir	108	9.00E‐25
0626		3	rRNA processing protein	335	A. fumigatus	an	vir	140	1.00E‐35
1394	1		N‐Myristoyltransferase	19	C. neoformans	an	vir	273	2.00E‐74
4690		1	Prolyl isomerase	436	C. neoformans	an	vir	143	6.00E‐36
2737		1	Glucosylceramide synthase	693	C. neoformans	an	path	110	1.00E‐25

^* A. brassicicola, Alternaria brassicicola; A. fumigatus, Aspergillus fumigatus; B. cinerea, Botrytis cinerea; C. carbonum, Cochliobolus carbonum; C. albicans, Candida albicans; C. nicotianae, Cercospora nicotianae; C. fulvum, Cladosporium fulvum; C. purpurea, Claviceps purpurea; C. posadasii, Coccidioides posadasii; C. heterostrophus, Cochliobolus heterostrophus; C. gloeosporioides, Colletotrichum gloeosporioides; C. graminicola, Colletotrichum graminicola; C. lagenarium, Colletotrichum lagenarium; C. lindemuthianum, Colletotrichum lindemuthianum; C. parasitica, Cryphonectria parasitica; C. neoformans, Cryptococcus neoformans; F. graminearum, Fusarium graminearum; F. oxysporum, Fusarium oxysporum; G. moniliformis, Gibberella moniliformis; L. maculans, Leptosphaeria maculans; M. grisea, Magnaporthe grisea; S. cerevisiae, Saccharomyces cerevisiae; S. nodorum, Stagonospora nodorum; T. virens, Trichoderma virens; U. maydis, Ustilago maydis; W. dermatitidis, Wangiella dermatitidis.

^† an, animal; gr, grass; pl, other plants.

^‡ effec, effector; hypvir, hypervirulence; path, loss of pathogenicity; unaff, unaffected pathogenicity; vir, reduced virulence.

^§ Sum of hosts > #PHI, pathogen(s) tested on different hosts; sum of hosts < #PHI, same gene with more than one PHI ID.

^¶ Arranged in descending order of occurring frequency.

^** Claviceps purpurea gene not yet in PHI‐base.

^†† EST identical with already cloned C. purpurea gene.

^‡‡ EST related to already cloned C. purpurea gene.

Table 5.

Expressed sequence tag (EST) contigs with homology to proteins involved in programmed cell death. The number and origin of the EST clones assembled in each contig are indicated: ax (ip) seqs, axenic (in planta) sequences.

Contig	ax seqs	ip seqs	Putative function	Identification†
0622	1	1	Metacaspase	blastx EST contig → nr NCBI
0639	1	1	Autophagy protein Atg20	blastx EST contig → nr NCBI
1509	1		Vacuolar fusion protein mon‐1	tblastn autophagy/apoptosis proteins (http://www.uniprot.org) → EST contigs
2231		1	Autophagy protein Atg15	tblastn Mg/Nz/Gz proteins (Table 2, Meijer et al., 2007) → EST contigs†
2460		1	Autophagy protein Atg29	tblastn Mg/Nz/Gz proteins (Table 2, Meijer et al., 2007) → EST contigs†
2489		1	Bax inhibitor family protein	blastx EST contig → nr NCBI
2542		1	Autophagy protein Atg9	tblastn Mg/Nz/Gz proteins (Table 2, Meijer et al., 2007) → EST contigs†
2652		1	Coat assembly protein SEC16	tblastn autophagy/apoptosis proteins (http://www.uniprot.org) → EST contigs
3115		1	Autophagy protein Atg8*	blastx EST contig → nr NCBI
3128		1	Protein‐vacuolar targeting protein Atg18	blastx EST contig → nr NCBI
3309		1	Autophagy protein Atg8*	blastx EST contig → nr NCBI
4677		1	Autophagy protein Atg13	blastx EST contig → nr NCBI

^* Contig 3315 contains introns, Contig 3309 does not.

^† Gz, Gibberella zeae; Mg, Magnaporthe grisea; Nc, Neurospora crassa.

Of the 127 EST contigs found (Table 4), 19 originate from the ax bank, 28 from both the ax and the ip bank, and 80 from the ip bank. Given that there are about three times more ip clones as ax clones, the frequency of hits among the ip contigs seems higher than expected. Table 4 shows the genes worth testing for roles in pathogenicity/virulence of Claviceps, yet functional analysis via knockout is necessary to validate the true pathogenicity/virulence genes.

A comparative analysis of EST collections between ascomycete pathogens and non‐pathogens showed that ESTs from certain functional categories are more highly represented in pathogenic than non‐pathogenic species: for example, ESTs concerning cell rescue, defence, death and ageing (Soanes and Talbot, 2006). This category includes ESTs involved in PCD. As, in filamentous fungi, PCD is involved in development, differentiation and sometimes also pathogenicity (for example, in M. grisea: Liu et al., 2007; Veneault‐Fourrey et al., 2006; in Colletotrichum gloeosporioides: Barhoom and Sharon, 2007; review: Ramsdale, 2008), we looked for potential PCD ESTs in our libraries (Table 5). We found 10, most of which are solely expressed in planta. Two are of particular interest: putative autophagy proteins Atg8 and Atg13. In yeast, Atg13 is linked to nutrient sensing via the protein kinase Tor (Meijer et al., 2007). During infection, C. purpurea displays targeted growth towards the rachilla, which is connected to the phloem and provides an endless supply of sugars. Atg8 in M. grisea is required for pathogenicity. Infection requires high turgor pressure in appressoria, which relies on autophagic PCD of the germinating spores (Veneault‐Fourrey et al., 2006). ‘Classical’ appressorium formation was not observed in Claviceps, but turgor cannot yet be ruled out as a factor in the penetration process.

Genes involved in nutrient uptake were of special interest. Some biotrophs, for example Uromyces, produce haustoria as nutrient‐absorbing organs, but these are not observed in Claviceps, which develops intracellular hyphae that may have haustorium function (Tenberge et al., 1996). Although we did not make a library mainly isolated from cells at the plant–fungus interface, the plant–fungus interface was part of our collected tissue, and it was instructive to compare our ESTs with highly expressed haustorial genes in the well‐investigated interaction of the biotroph Uromyces fabae with Vicia faba (Jakupovičet al., 2006) (see Table 6). Of the 35 Ur. fabae haustorial genes whose expression increased five times or more in rust‐infected leaves vs. germinated uredospores, 14 (40%) showed a significant hit to a Claviceps EST contig. A similar search with haustorial EST clones from Blumeria graminis (an obligate grass biotroph) that are up‐regulated five‐fold in planta yielded no matches to Claviceps ESTs or to the Uromyces up‐regulated haustorial genes (data not shown). This could be a reflection of a different host colonization mode. Blumeria has epicuticular hyphae with intracellular haustoria, whereas both Uromyces and Claviceps initially grow with intracellular hyphae and later with intercellular hyphae. The identification of genes encoding a sugar transporter and a mannitol dehydrogenase—via a hit to the haustorial rust genes—indicates the importance of sugars and sugar polyols in ergot disease. Sugar and sugar polyols are constituents of the honeydew produced in phase II of the infection. The enzyme mannitol dehydrogenase may have a secondary role as a defence molecule against ROS produced by the host (Solomon et al., 2007). In addition, manganese superoxide dismutase may have an antioxidative role. Glutamine synthetase could play a role in nitrogen metabolism, and a sulphur regulatory protein cys‐3 a role in sulphur metabolism. Two contigs were found which code for proteins involved in protein folding. Contig 1031 may encode a peptidyl–prolyl cis–trans‐isomerase known to be a folding catalyst and Contig 0345 a mitochondrial HSP60, which participates in the folding of mitochondrial proteins. It would be interesting to determine the actual expression levels of these Claviceps ESTs. It might be possible to use them (via fusion to fluorescent proteins) to investigate whether intracellular ergot hyphae have haustorium function.

Table 6.

Claviceps expressed sequence tag (EST) contigs with significant homology to haustorial Uromyces fabae genes highly expressed in rust‐infected leaves. The number and origin of the EST clones assembled in each contig are indicated: ax (ip) seqs, axenic (in planta) sequences.

Contig	ax seqs	ip seqs	Putative function	Match with U. fabae haustorium: Clone identifier (Accession)*	HSP	E‐value	Expression in U. fabae †
1298	7	6	Thiamine biosynthesis protein	THI1 (O00057)	468	1.00E‐133	53.9
1162	1	12	Mannitol dehydrogenase	MAD1 (O00058)	154	6.00E‐39	49.5
0359	4	5	Glutamine synthetase	Uf101 (DR010279)	146	1.00E‐62	24.7
0200	1	5	Manganese superoxide dismutase	Uf058 (DR010236)	67	2.00E‐27	17.8
0777	9	5	Thiamine biosynthesis protein	THI2p (CAB59856)	375	1.00E‐105	13.5
0457	3	1	MFS monosaccharide transporter	HXT1 (CAC41332)	155	7.00E‐39	10.1
3298		1	Regulatory protein cys‐3	Uf053 (DR010231)	96	7.00E‐21	9.9
0527	3	5	60s ribosomal protein L7a	Uf478 (DR010656)	209	3.00E‐55	9.4
1031	13	7	Peptidyl–prolyl cis–trans‐isomerase	PIG28 (AAB39880)	219	6.00E‐59	8.7
3116		1	Cytochrome P450 monooxygenase	PIG16 (AAB39881)	91	1.00E‐19	7.6
0037	1	1	40S ribosomal protein S28	Uf150 (DR010328)	109	3.00E‐25	7.3
0825	1	2	60S ribosomal protein L44	Uf371 (DR010549)	238	3.00E‐64	6.5
0345	3		Mitochondrial heat shock protein 60	Uf130 (DR010308)	138	1.00E‐57	5.8
0724	8	3	S‐phase specific ribosomal protein	Uf406 (DR010584)	188	3.00E‐49	5

^* tblastx hit for nucleotide sequences, tblastn hit for protein sequences. (Nucleotide sequences start with DR0 . . . All others are protein sequences.)

^† Ratio of expression in infected leaves vs. expression in germinated uredospores.

Comparison of Claviceps with other flower pathogens

Flower pathogens benefit from easy entry into the host through thin‐walled petals and stigmas, readily available nutrients (for example, nectar) and efficient spore dispersal via co‐opted avenues that evolved for the distribution of pollen and seeds. However, the phenology of flower diseases is critical, as anthesis occurs in a narrow window of time, is short lived and is poised for rapid responses to environmental stresses. Three groups of flower pathogen have been described: (i) unspecialized pathogens, which generally do not produce survival structures on flowers; (ii) specialized pathogens, which infect ovaries through the stigma–style pathway and often have an obligate sexual cycle on the flower, and replace the host's inflorescence with resting structures; and (iii) specialized pathogens, which gain entry via the apical meristem and often grow for a long time endophytically in the host (Ngugi and Scherm, 2006).

Claviceps purpurea belongs to group (ii). The necrotroph flower pathogen F. graminearum belongs to group (i) and is, like Claviceps, a member of the order Hypocrales. Even their hosts, barley and rye, are both members of the Triticaceae. Another member of group (ii) is the basidiomycete U. maydis, which has a biology and lifestyle similar to that of Claviceps (O'Connell and Panstruga, 2006; Ngugi and Scherm, 2006): for example, both are biotrophs and lack haustoria.

Fusarium graminearum and U. maydis are well‐characterized organisms. Their genomes have been sequenced and annotated with respect to potential pathogenicity genes. A collection of enriched PHI‐base sequences from both organisms was compared with the Claviceps ESTs and with each other. The F. graminearum collection also included about 400 sequences that are expressed solely during barley infection and located in regions enriched in single nucleotide polymorphisms. About 30% of these sequences are predicted to be part of the secretome (Cuomo et al., 2007; Güldener et al., 2006). This collection also includes the sequences described in Paper et al. (2007) as proteins identified at high reliability in infected wheat heads and genes, and found by Stephens et al. (2008) to be significantly up‐regulated in both crown rot and head blight of wheat. The U. maydis collection also includes sequences published in the genome paper (Kämper et al., 2006), as well as those predicted to be secreted proteins with or without enzymatic function (Mueller et al., 2008)—assuming that secreted proteins are good candidates for being involved in the disease process. Many of the Ustilago sequences appear to be specific to Ustilago.

Results of the comparisons are found in Table S2 (see Supporting Information). The interpretation of the data is constrained as each dataset was collected under different conditions. That aside, the following observations were made.

• 1No match to a U. maydis‐specific sequence was found, which is not unexpected as most pathogens probably have both specific and general virulence factors.
• 2Most matches were found between C. purpurea/F. graminearum, perhaps because of their close taxonomic relationship. Yet, despite belonging to two different taxonomic groups, more matches were noted between U. maydis and C. purpurea than between U. maydis and F. graminearum. This may reflect the similar lifestyles of corn smut pathogen and the ergot pathogen.
• 3In all three fungi, signalling components and cell wall‐degrading enzymes were commonly found, which is reasonable for plant pathogens in general. Yet, in each set of pairwise comparisons, different genes/proteins were prevalent: between U. maydis and F. graminearum, additional cell wall‐degrading enzymes were found; between F. graminearum and C. purpurea, transporters for specific molecules, oxidoreductases/dehydrogenases and proteinases/peptidases were observed; and between U. maydis and C. purpurea, molecules involved in cell cycle control, transport (different from the F. graminearum/C. purpurea matches) and protein assembly/folding were apparent. The latter observation may reflect the fine‐tuning necessary in a biotrophic host–pathogen interaction.

Intriguingly, within the corn smut/ergot matches, a pathogen‐related protein with similarity to the PR1 class of plant proteins was found. PR proteins are produced by plants to ward off attacking fungi. In particular, proteins similar to classes PR1 and PR5 have also been found recently in animals and fungi, but they have not been well characterized as yet. Mondego et al. (2008) have suggested that such proteins may limit microbial competition for the hemiobiotrophic cacao pathogen Moniliophthora perniciosa during progression of witches' broom disease. The production of a PR1 protein by a pathogen may also be part of a strategy to overcome plant defences. Perhaps the plant is mislead into prematurely sensing that the level of PR1 protein is sufficient for its protection and reacts too late to avoid a successful attack. As in many situations, the timing of the host's and pathogen's actions may be decisive for the outcome of the infection.

EXPERIMENTAL PROCEDURES

Fungal and plant material, and culture conditions for cDNA libraries

For the axenic cDNA library, C. purpurea strain 20.1 was grown for 3 days at 28 °C and 180 r.p.m. in Mantle medium (Mantle and Nisbet, 1976). Mycelium was harvested by centrifugation and lyophilized before RNA preparation. For in planta cDNA libraries, flowers of S. cereale cultivar Lo37‐PxLo55‐N (cytoplasmic male sterile) (KWS Lochow GmbH, Bergen, Germany) were inoculated with 5 µL of a suspension containing 2 × 10⁶ conidia/mL. All ovaries (except for bases and tops) from inoculated spikes were harvested, submersed in liquid nitrogen and lyophilized before RNA preparation. For each day, from day 1 up to day 20 post‐inoculation (1–20 dpi), ovaries from three rye spikes were collected.

cDNA library constructions

Extraction of total RNA from both infected rye ovaries and fungal mycelium was performed with the ‘RNAgents® Total RNA Isolation System’ from Promega, Mannheim, Germany, according to the instructions of the supplier. Total RNA from ovaries at 1–5 dpi was pooled for an early and from ovaries at 6–20 dpi for a late in planta cDNA library. mRNA was isolated from all three total RNAs with an ‘Oligotex^TM mRNA Mini Kit’ from Qiagen, Hilden, Germany, according to the manufacturer's instructions. Three primary cDNA libraries (axenic, early and late in planta) were produced with a ‘pBluescript II cDNA Library Construction Kit’ from Stratagene, La Jolla, CA, USA, following the instruction manual.

Clone isolation

About 100 clones of each cDNA library were plated and their pBluescript II inserts were sequenced from the 5′ cDNA end as a quality check. Two of the three libraries passed the test. The early in planta library failed the test, because about 25% of the cDNA clones contained only the vector and another 25% had inserts that resisted sequencing (possibly as a result of a loss of the sequencing primer binding site). From the remaining ax and ip cDNA libraries, clones were plated; each was manually transferred into freezing medium (3 : 1 ratio of 2 × Luria–Bertani medium and glycerol), grown overnight at 37 °C and manually aliquoted into three parallel 96‐well microtitre plates. The ax library was stored in 26 96‐well microtitre plates (clones ax_001_A01 to ax_026_H12) and the late ip library in 78 96‐well microtitre plates (clones ip_027_A01 to ip_104_H12). Two plates of one set were stored at −70 °C. The clones of one plate were sequenced.

DNA sequencing

The 5′ ends of each cDNA clone were sequenced via Sanger sequencing using an Applied Biosystems 3730 capillary DNA sequencer by GreenTec GmbH (now Phytowelt GreenTechnologies GmbH, Nettetal/Cologne, Germany). Sequences were delivered in ab1 and fasta format.

Processing of sequences

Raw DNA sequences were trimmed with LUCY2 (Li and Chou, 2004). The output was screened for sequences containing putative internal linker sequences, indicating that two different cDNAs may have been ligated together. All potential coupled cDNAs were manually separated into two parts for the time of assembly and clustering (and assigning plant or fungal origin). These sequences were labelled afterwards with a ‘j’, for example ax_007jD12. Resulting sequences shorter than 100 bp were deleted from the project. About 1500 sequences rejected by LUCY2 were manually screened. All those longer than 100 bp and not consisting of short, monomeric stretches of Gs, As, Ts and Cs were accepted. All 343 included files were vector clipped, quality trimmed and indicated with an ‘m’ in the label, for example ax_006mF03.

The sequences were clustered with tgicl (Pertea et al., 2003). To find appropriate parameters, a test set of 29 published C. purpurea genes was clustered under default conditions (–p 94 –l 30 –v 30) and under conditions (–p 88 –l 21 –v 900) which resulted in clusters of very closely related genes and also clusters formed by spliced and unspliced versions of each gene. As ESTs are usually shorter than their corresponding genes, the final parameters chosen were –p 90 –l 30 –v 30.

SeqMan Pro (Lasergene 7.1 by DNASTAR, Madison, WI, USA) was used to assemble the EST sequences into contigs. Under default conditions, the assembly contained contigs with sequences in which large gaps (40–130 bp) had been introduced into sequences by the program to fit the sequence into the contig. Therefore, the assembly conditions were adjusted to Match Size = 50, Match Spacing = 50, Maximum Mismatch End Bases = 0, Use Repeat Handling = Yes, Fixed = 4 and Match Repeat Percentage = 150. To find at least a good part of the falsely joined sequences, contigs were checked for suspiciously small overlaps or long overhangs in the strategy view. Such contigs were subjected to a blastx search against nr of NCBI (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi?PAGE=Translations&PROGRAM=blastx&BLAST_PROGRAMS=blastx&PAGE_TYPE=BlastSearch&SHOW_DEFAULTS=on). If the search revealed two regions with fundamentally different matches, the sequences of such a contig were split into two contigs. Table S1 shows which EST clones belong to which contig and cluster.

Sequence analysis

GO annotation (comparing ax and ip libraries)

GO terms of two aspects (‘Molecular function’ and ‘Biological process’) were assigned to all ax and ip ESTs via GOPET (‐CONFIDENCE = 80) (Vinayagam et al., 2006). For each aspect, the terms found were counted separately for ax ESTs and ip ESTs with the GO terms classifications counter (Hu et al., 2007; http://www.animalgenome.org/bioinfo/tools/countgo/). The GO term counts for the ax and ip libraries were checked by the method of Audic and Claverie (1997) for significantly different counts (significance level = 0.01).

Assignment of plant or fungal origin for ip ESTs

All ax ESTs were organized into a fungal databank. A blastn search of all ip ESTs was performed against the ax databank. This was accomplished using seqtools (Dr Søren Rasmussen, H. Lundbeck A/S, Copenhagen, Denmark, http://www.seqtools.dk), which was also employed to build additional databanks and to perform blast searches with these databanks (against one of the self‐assembled or NCBI databanks). All hits with an E‐value ≤ 9E‐100 were considered to be of putative fungal origin. Similarly, databanks were assembled from genomes and EST libraries of organisms closely related to C. purpurea or S. cereale, as neither genome has yet been sequenced:

• •  Epichloë festucae genome (no databank built, but use of blastn offered in E. festucae Genome project webpage; last modified October 2000, http://www.endophyte.uky.edu/index.php);
• • genomic Clavicipitaceae and Triticeae sequences ≥100 bp from NCBI (http://www.ncbi.nlm.nih.gov/, downloaded July 2007);
• •  F. graminearum genome (http://www.broad.mit.edu/annotation/genome/fusarium_graminearum/Home.html, March 2007);
• •  F. verticillioides, M. grisea cDNA sequences (http://compbio.dfci.harvard.edu/tgi/fungi.html; release 7.0 of Fusarium and release 5.0 of Magnaporthe libraries #0QV, #A36–39, #A3A, #CN6, #F3J, #F3P);
• •  S. cereale, T. aestivum, H. vulgare unique transcript sequences (http://www.plantgdb.org/download/download.php?dir=/Sequence/ESTcontig, version 157a for rye and barley, version 153a for wheat);
• •  O. sativa cDNA sequences (http://compbio.dfci.harvard.edu/tgi/cgi‐bin/tgi/cat_download.pl?db=ricest, #GCH—a normalized whole‐life‐cycle cDNA library of rice, release 17.0).

All blastn hits against a fungal/plant databank with an E‐value ≤ 9E‐100 were considered to be of putative fungal/plant origin. If there were hits of one EST with both fungal and plant databanks, the lowest E‐value/highest HSP score were used as the decisive criteria. With all ESTs that could not be sorted, a blastx search against the nr databank of NCBI and a blastn search against non‐human, non‐mouse ESTs (EST_other) of NCBI were performed. The chosen criteria discriminate against smaller sequences (≤250 bp), whose blastn searches will not result in such low E‐values. Yet, empirically, we found that, for a given average sequence (GC content between 50 and 60%), there is a linear relationship between sequence length and HSP score, if it is blasted to itself (query length × 1.98 ≈ HSP score). There is also a linear relationship if the query sequence has 90% similarity with the sequence it is blasted against (query length × 1.17 ≈ HSP score_90%). Therefore, small EST sequences were considered to be of putative fungal/plant origin when they achieved an HSP score higher than a calculated HSP score_90%. Finally, all putative plant ip clones were traced back to their contig. There was no occurrence of ‘mixed’ contigs. Table S1 shows the (putative) origin of each EST clone.

For comparison with the putative plant EST contigs, two EST banks were assembled from the TGI plant Gene Index Project (http://compbio.dfci.harvard.edu/tgi/plant.html): (i) a flower EST bank (59 273 sequences) comprised libraries 5474, #8GO, 5341, 7313 from H. vulgare, #DOH, #G8E, #IJA, #IJH, #ILF, #ILG, #IMK, #IOP, OS31, #IKR, #IKS, #IKT, #ILI, #ILJ, #ILS from O. sativa, #E14 from Sorghum bicolor, #5CC, #DF2 from S. cereale, 5459, #DEP, 5603 from T. aestivum, and ZM19 from Zea mays; (ii) an infected flower EST bank (11 939 sequences) comprised library #9JH from H. vulgare and #9HS, #9O0, #AS6, #AS7, #AS8, #DE6, 5584 from T. aestivum. Significant hits were those with an E‐value ≤ 1E‐38.

Comparison with protein sequences in PHI‐base (version 3.1, http://www.phi‐base.org/)

Protein sequences collected in PHI‐base were downloaded, arranged in a databank (together with four C. purpurea genes not yet part of PHI‐base) and run in a tblastn search against the EST contigs of C. purpurea (e < −15). A hit was rejected if the contig had been classified previously to be of putative plant origin. If there was more than one PHI‐base sequence originating from the same organism with a significant hit to the same contig, only the hit with the lowest E‐value was listed in Table 4. In addition, each query and hit pair of this search was subjected to a blastp/blastx search to the nr databank of NCBI. If the result of these searches did not correspond, the hit was rejected after closer manual inspection of the hits.

Comparison with U. fabae genes highly expressed in haustoria

The comparison was similar to the comparisons with PHI‐base sequences and secretome sequences. Selected for comparison were U. fabae genes published in Jakupovičet al. (2006), showing five‐fold or greater up‐regulation in rust‐infected leaves vs. germinated uredospores. The accession numbers given in the publication refer to both nucleotide and protein sequences. The nucleotide sequences were compared with the C. purpurea EST contigs via tblastx, the protein sequences via tblastn.

Comparison with enriched PHI‐base protein sequences of U. maydis or F. graminearum only

Additional U. maydis protein sequences were downloaded and subselected from the MIPS Ustilago ftp site (ftp://ftpmips.gsf.de/ustilago/, Umaydis_valid_orf_prot_140408.fas) according to the genes listed in supplementary Tables S1, S2 and S3 of the recently published secretome paper of Ustilago (Mueller et al., 2008).

Additional F. graminearum sequences correspond to those listed as ‘in planta only’ in supplementary Table D in the publication of Güldener et al. (2006). The nucleotide sequences were downloaded from the PLEXdb site http://www.plexdb.org/modules/tools/get_seq.php (Choose microarray ‘Fusarium’). To mass translate these nucleotide sequences correctly into protein sequences, they were blasted against Fgraminearum_valid_orf_prot_2007‐05‐10 from the MIPS Fusarium ftp site (ftp://ftpmips.gsf.de/fusarium/). Furthermore, all sequences listed in Table 2 (Paper et al., 2007) and in Table S2 (Stephens et al., 2008) were included in this F. graminearum sequence collection.

Searches were performed similar to that with PHI‐base sequences. If there were two or more sequences with a significant hit to the same contig, only the best hit (that with the lowest E‐value) was listed in Table S2. In addition, hits were selected only if the result of a blast of sequence 1 from organism 1 against sequence 2 of organism 2 gave a result similar to the blast of sequence 2 from organism 2 against sequence 1 of organism 1 (reciprocal search).

Supporting information

Table S1 Overview of expressed sequence tag (EST) clones. Listed are the contig and cluster association, the respective EMBL accession number and the (putative) origin. n.d., not determinable; put, putative.

Table S2 Comparison of (selected) sequences from three flower pathogens.

Please note: Wiley‐Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Supporting info item

Supporting info item