Roles of the Glyoxylate and Methylcitrate Cycles in Sexual Development and Virulence in the Cereal Pathogen Gibberella zeae

Abstract

The glyoxylate and methylcitrate cycles are involved in the metabolism of two- or three-carbon compounds in fungi. To elucidate the role(s) of these pathways in Gibberella zeae, which causes head blight in cereal crops, we focused on the functions of G. zeae orthologs (GzICL1 and GzMCL1) of the genes that encode isocitrate lyase (ICL) and methylisocitrate lyase (MCL), respectively, key enzymes in each cycle. The deletion of GzICL1 (ΔGzICL1) caused defects in growth on acetate and in perithecium (sexual fruiting body) formation but not in virulence on barley and wheat, indicating that GzICL1 acts as the ICL of the glyoxylate cycle and is essential for self-fertility in G. zeae. In contrast, the ΔGzMCL1 strains failed to grow on propionate but exhibited no major changes in other traits, suggesting that GzMCL1 is required for the methylcitrate cycle in G. zeae. Interestingly, double deletion of both GzICL1 and GzMCL1 caused significantly reduced virulence on host plants, indicating that both GzICL1 and GzMCL1 have redundant functions for plant infection in G. zeae. Thus, both GzICL1 and GzMCL1 may play important roles in determining major mycological and pathological traits of G. zeae by participating in different metabolic pathways for the use of fatty acids.

During the infection process, pathogenic fungi usually encounter nutrient deprivation in the host before gaining access to sufficient nutrients for successful colonization of the living tissue. To cope with a nutrient-limited environment, fungal pathogens seem to rely mostly on fatty acid metabolism for both energy supply and biosynthesis of essential molecules (29). The ability of fungi to use fatty acids as a carbon source for growth is based on the glyoxylate cycle. Fungal pathogens have been proposed to employ the glyoxylate bypass for the use of acetyl coenzyme A (CoA) units produced by the β-oxidation of even-chain-length fatty acids, probably available from host cell membranes or the lipid reservoir inside the fungal spore (7, 12, 20, 27, 28, 41, 44, 46). Recent studies suggest that the glyoxylate pathway plays an important role in fungal virulence toward both plant and animal hosts (12, 20, 27, 44, 46). The key enzymes of the glyoxylate pathway, such as isocitrate lyase (ICL), which catalyzes the cleavage of isocitrate to glyoxylate and succinate, and malate synthase, which mediates the condensation of acetyl-CoA and glyoxylate into malate, are strongly induced within the host (16, 27, 41, 44). Moreover, disruption of genes encoding either of these enzymes causes severely reduced virulence of fungal phytopathogens, including Leptosphaeria maculans (20), Magnaporthe grisea (46), Stagonospora nodorum (44), and Colletotrichum lagenarium (2), and the animal pathogen Candida albicans (27). In contrast, these glyoxylate cycle enzymes have been known to be dispensable in invasive aspergillosis caused by Aspergillus fumigatus (38, 43).

During fatty acid and amino acid catabolism by fungi, propionyl-CoA can be generated along with acetyl-CoA, particularly from the breakdown of odd-chain-length fatty acids or of the amino acids valine, isoleucine, and methionine (14). Therefore, fungal pathogens may need to use or remove propionyl-CoA during the infection process because it is toxic to fungi. In fungi, propionyl-CoA is metabolized via the methylcitrate cycle, in which propionyl-CoA is oxidized to pyruvate in four enzymatic steps (4, 5, 6, 19, 30, 31, 40, 49, 50). Recently, the importance of the methylcitrate cycle in fungal virulence was demonstrated in A. fumigatus: a mutant defective in methylcitrate synthase, the first enzyme of this cycle, displayed attenuated virulence in mice and insects (19, 31). However, the role of methylisocitrate lyase (MCL), which catalyzes the last reaction in the methylcitrate cycle (i.e., the cleavage of methylisocitrate into pyruvate and succinate) in fungal virulence, has not been determined, although deletion of the MCL gene inhibits hyphal growth and conidiation in Aspergillus nidulans (4). The protein sequences of several fungal MCLs show high similarity to fungal ICLs of the glyoxylate cycle (4, 30). In the pathogenic bacterium Mycobacterium tuberculosis, the methylcitrate cycle, only when working together with the glyoxylate cycle, is involved in virulence as well as fatty acid metabolism and intracellular growth (34, 35).

Here, we focused on the roles of these two cycles during disease development caused by the devastating cereal pathogen Gibberella zeae (anamorph: Fusarium graminearum). G. zeae is a ubiquitously distributed ascomycete fungus that causes major disease in cereal crops such as corn, wheat, barley, and rice (33). Severe epidemics of these diseases result in serious economic consequences due to yield losses and contamination by fungal mycotoxins (32, 33). Wind-disseminated sexual spores (ascospores), which are produced in perithecia formed on plant debris, can infect plant spikes during anthesis (13, 39, 45). Detailed studies of the G. zeae infection process on wheat and barley heads have shown that fungal hyphae on the inner surfaces of the spike penetrate epicarp cells through pits or pores and grow into the caryopses through the pericarp (21). Thus, the glyoxylate cycle, either alone or in conjunction with the methylcitrate cycle, is likely employed by G. zeae during the infection process, as in other fungus-plant interactions (20, 46). G. zeae genome searches have identified orthologs of fungal ICL and MCL genes, designated GzICL1 and GzMCL1, respectively. Here, we performed functional analyses of these genes to provide new insight into their importance in lipid metabolism during the G. zeae infection process in host plants.

MATERIALS AND METHODS

Strains, culture conditions, and fungal transformation.

Gibberella zeae strain Z03643, a lineage 7 and self-fertile strain (37), was obtained from Robert L. Bowden, U.S. Department of Agriculture, Manhattan, KS. GzΔMAT1 is a self-sterile strain derived from a mat1-1-deleted Geneticin-resistant G. zeae strain (T39ΔM1-1) (25), which is a mat1-1-deleted progeny carrying no gen from the outcross between T39ΔM1-1 and Z03643 (S.-H. Lee et al., unpublished data). Fungal strains from 25% glycerol stock cultures stored at −80°C were maintained on potato dextrose agar (PDA; Difco Laboratories, Detroit, MI). For both perithecial formation and total RNA extraction, the strains were grown on carrot agar plates (26). For vegetative growth and genomic DNA extraction, the strains were grown on either PDA or complete medium (26) or on basal medium (minimal medium without sucrose) (26). Escherichia coli strains were grown on Luria-Bertani agar or in liquid medium supplemented with 75 μg/ml ampicillin. Fungal transformation was performed as previously described (25).

Vector construction.

To construct the plasmid used for targeted deletion of GzICL1, we employed an inverse PCR strategy. A 4.0-kb DNA carrying the inverse PCR entire GzICL1 open reading frame (ORF) (plus 1,016-bp 5′ and 801-bp 3′ flanks) was amplified from the genomic DNA of Z03643 using GzICL1-BglII-F and GzICL1-BglII-R primers (see Table S1 in the supplemental material) carrying the recognition sites for BglII at the 5′ ends. The PCR product was digested with BglII, self-ligated, and used as a template for inverse PCR with primers GzICL1-delF and GzICL1-delR (see Table S1 in the supplemental material). An 1,817-bp fragment carrying both the 5′ and 3′ flanks of GzICL1, but not the ORF, was inserted into the EcoRV site of the pCB1004 vector (kindly provided by Seogchan Kang, Penn State University, University Park, PA) carrying hygB as a selectable marker. For the deletion of the GzICL1 gene, the resulting 6,617-bp plasmid designated pCBGzICL1, was linearized with BglII and introduced into G. zeae protoplasts.

A transforming DNA fragment used for the deletion of GzMCL1 was constructed using the double joint PCR method (48). DNA fragments corresponding to 5′ (1,633 bp) and 3′ (1,672 bp) regions of the GzMCL1 ORF were amplified from the genomic DNA of Z03643 using the primer pairs GzMCL1-F/GzMCL1-FGT and GzMCL1-RGT/GzMCL1-R (see Table S1 in the supplemental material), respectively. A 1.8-kb fragment containing the gen gene was amplified from the vector pII99 (36) using primers Gen-F and Gen-R. Three amplicons (the 5′-flanking region of GzMCL1, the gen cassette, and the 3′-flanking region of GzMCL1) were mixed in a 1:2:1 molar ratio and used as a template for a second round of PCR with a new primer pair (GzMCL1-NF and GzMCL1-NR; see Table S1 in the supplemental material), resulting in a 3,903-bp fragment of the fusion PCR product. Following purification, the final PCR products were mixed with fungal protoplasts for use in transformation for the deletion of GzMCL1.

For complementation analyses, intact copies of GzICL1 and GzMCL1 including the native promoter and terminator were amplified from genomic DNA of the Z03643 strain using primer pairs GzICL1-NF/GzICL1-NR and GzMCL1-NF/GzMCL1-NR, respectively, and that of A. nidulans acuD was amplified from A. nidulans FGSC A4 using the primer pair AnICL1-NF/AnICL1-NR (see Table S1 in the supplemental material). The amplified GzICL1, or acuD, product was directly added into the ΔGzICL1 protoplasts along with the pII99 vector carrying gen as a fungal selectable marker; the GzMCL1 PCR product was introduced into the ΔGzMCL1 protoplasts along with pUCH1 carrying hygB.

Nucleic acid manipulations.

Total RNA for RNA blots and reverse transcriptase (RT)-PCR analysis was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Fungal genomic DNA was extracted using the cetyltrimethylammonium bromide protocol, as described previously (26). Plasmid DNA was purified using a plasmid DNA purification kit (Intron Biotechnology, Seoul, Korea). Standard procedures were used for gel electrophoresis, restriction enzyme digestion, ligation, and blotting (42). RT-PCR was performed using the AccuPowerRT/PCR PreMix (Bioneer, Daejeon, Korea) according to the manufacturer's protocol. Quantitative real-time PCR was performed with the SYBR green super mix (Bio-Rad, Hercules, CA) and a 7500 real-time PCR system (Applied Biosystems, Foster, CA). The PCRs were repeated three times with three replicates per run. The G. zeae gene encoding elongation factor 1-β (EF1b; FGSG_01008.3) was used as an endogenous control for normalization. The cycle threshold (C_T) value of EF1b was subtracted from that of PKS12 (FGSG_02324.3) (23) encoding a polyketide synthase responsible for the production of the G. zeae red pigment (aurofusarin) to obtain a ΔC_T value. The ΔC_T value of an arbitrary calibrator was subtracted from the ΔC_T value of each sample to obtain a ΔΔC_T value. The PKS12 expression level relative to the calibrator was expressed as 2^−ΔΔCT.

Sexual crosses.

Both a self-cross and an outcross of each G. zeae strain were performed in carrot medium, as described previously (25, 26). A self-cross was initiated by placing mycelial agar blocks of a fungal strain at the centers of carrot agar plates and incubating at 25°C. After 10 days (during vegetative growth), the aerial mycelia that had grown on carrot agar were then removed by scraping the culture with a conidial suspension (10⁵ conidia/ml in 2.5% Tween 60) of the same strain, and the plates were incubated for an additional 10 to 14 days for perithecial induction at 25°C under a mixture of fluorescent cool white and black lights with a 12-h photoperiod. An outcross was initiated by placing mycelial agar blocks of the mat1-1-deleted strain (GzΔMAT1) of G. zeae on carrot agar plates. After 10 days at 25°C, conidial suspensions (1 × 10⁵ conidia/ml) of the fungal strains were applied to the mycelia of GzΔMAT1, and the plates were incubated as described above. Each outcross was set up in 10 carrot agar plates. More than 50 perithecia were randomly picked up on the mating plates of each cross, and two ascospores were isolated from each perithecium for genetic analysis.

Virulence tests.

Virulence tests were performed using spikelets of the barley cultivar SangRok and the wheat cultivar Eunpamil, both of which are very susceptible to head blight. For fungal inoculation, two different methods were employed. G. zeae conidial suspensions containing 1 × 10⁶ conidia were prepared in CMC (carboxymethylcellulose) medium (8) and sprayed evenly onto barley and wheat heads, as described previously (18) (spray inoculation method), or approximately 10 μl of the G. zeae spore suspension (1 × 10⁵ spores/ml) was injected into a basal spikelet of the wheat head at midanthesis (point inoculation method). For each treatment, six barley or wheat heads were inoculated, and inoculated plants were placed in a humidity chamber for 3 days and then transferred to a greenhouse under normal conditions until disease symptoms appeared. At 14 days postinoculation, spikes inoculated by the spray method were visually rated and the average severity of head blight was calculated as follows. The sum of infected spikelets for each symptom severity on a 0 to 4 scale (0, no symptoms; 1, <25% necrotic regions on a spikelet; 2, <50%; 3, <75%; 4, 100% necrosis and/or bleaching) divided by the total number of spikelets inoculated. All data were obtained from two biological replicates, and the Tukey test using SPSS 12.0 software (SPSS Inc., Chicago, IL) was performed to examine the significant differences (P < 0.05) of disease severity among the mean values of samples.

ICL and MCL assays.

Isocitrate lyase activity was determined using cell extracts following the method of Dixon and Kornberg (10). The cell extracts were prepared as described by De Lucas et al. (9) and directly assayed for enzyme activity. The amount of protein in each mutant was quantified using Bradford's method (3). The assay mixture consisted of 8 mM d,l-isocitric acid, 6 mM MgCl₂, 4 mM phenylhydrazine, 12 mM cysteine, and 62.5 mM potassium phosphate buffer (pH 7.0). The reaction was started by adding cell extract, followed by incubation for 5 min at 25°C. Formation of the enzymatic product was monitored at a wavelength of 324 nm using a spectrophotometer (Ultospec 4000; Pharmacia Biotech, Tokyo, Japan). For the methylisocitrate lyase activity assay, we used 2-methylisocitrate lactone, a kind gift from Matthias Brock, Leibniz Institute for Natural Product Research and Infection Biology, Hans Knoell Institute, Jena, Germany, as the enzyme substrate for measurement of MCL activity, as previously described (4). Average activities for each enzyme were obtained from three replicates.

RESULTS

Sequence similarity.

The GzICL1 and GzMCL1 genes were identified from the F. graminearum genome (http://www.broad.mit.edu/annotation/fungi/fusarium/index.html). The putative ORF of GzICL1 (FGSG_09896.3; annotated as an isocitrate lyase in the fungal genome database) is 1,893 bp long and contains four putative introns; the putative 1,800-bp ORF of GzMCL1 (FGSG_00176.3; annotated as another isocitrate lyase in the genome database) carries one putative intron. The deduced GzICL1 amino acid sequence exhibited significant similarity to other fungal ICLs, such as acu3 of Neurospora crassa (84% identity) (15), ICL1 of M. grisea (81% identity) (46), ICL of L. maculans (75% identity) (20), and acuD of A. nidulans (74% identity) (15). The putative polypeptide of GzMCL1 showed 69% identity to A. nidulans mclA (CAI65406) encoding MCL, as well as other hypothetical MCLs or ICLs (see Fig. S1 in the supplemental material). The putative GzMCL1 protein carried a predicted sequence motif (GFVLQLISLAGLH) specific for fungal MCLs (4). When we searched for a putative mitochondrial targeting sequence using the program MITOPROT (http://ihg.gsf.de/ihg/mitoprot.html), only the GzMCL1 protein was transported to mitochondria, with a probability of 0.9996. The two G. zeae genes, GzICL1 and GzMCL1, shared 43% identity with each other at the amino acid level.

Targeted gene deletions and complementation.

To determine the role(s) of GzICL1 and GzMCL1 in G. zeae, we deleted one or both genes from the genome of strain Z03643 (Fig. 1). First, the putative GzICL1 ORF was replaced with the vector pCBGzICL1 carrying the fungal selectable maker hygB via double homologous recombination between both the 5′ and 3′ regions of the GzICL1 ORF on the vector and the corresponding genomic regions (Fig. 1A). Genomic DNA of strains with the deletion of GzICL1 (designated ΔGzICL1) carried a 9.4-kb hybridizing band when digested with BglII, instead of the 6.5-kb band found in the wild-type Z03643 strain, suggesting that the 1.9-kb GzICL1 ORF had been deleted and replaced with the hygB gene (Fig. 1A). For the deletion of GzMCL1 (ΔGzMCL1), the ORF of GzMCL1 was replaced with the gen gene by double homologous recombination between the double joint PCR product and the fungal genome (Fig. 1B). The ΔGzMCL1 strains derived from Z03643 showed a single 5.0-kb hybridizing band on a blot of HindIII-digested genomic DNA instead of the 2.7-kb band in the wild-type Z03643 strain (Fig. 1B). For the double deletion of GzICL1 and GzMCL1 (designated ΔGzICL1 ΔGzMCL1), we removed the GzMCL1 gene from the genome of a ΔGzICL1 strain (TdGI1-1) using the same method described above. Double gene deletion in the ΔGzICL1 ΔGzMCL1 strains was confirmed by DNA gel blot analysis (Fig. 1B). The insertion of the GzICL1, acuD, or GzMCL1 ORF at the genomes of the complemented transformants was also confirmed by DNA gel blot hybridization (Fig. 1C).

FIG. 1.

Targeted deletion of GzICL1 (A) or GzMCL1 from the genome of wild-type Gibberella zeae strain Z03643 (B) and genetic complementation (C). (A and B) Left portions of each panel, deletion strategies. WT, genomic DNA of Z03643; ΔGzICL1, genomic DNA of the strain with GzICL1 deleted; ΔGzMCL1, genomic DNA of the strain with GzMCL1 deleted; B, BglII fragment; H, HindIII fragment; hygB, hygromycin B resistance gene; gen, geneticin resistance gene. The probes used for DNA gel blot hybridization are indicated by bars. Right portions of each panel, DNA gel blot hybridizations. Lane 1, Z03643; lane 2, a ΔGzICL1 strain; lane 3, a ΔGzMCL1 strain, lane 4, a double deletion ΔGzICL1 ΔGzMCL1 strain. (C) Lane 1, Z03643; lane 2, a ΔGzICL1 strain; lane 3, a ΔGzICL1::GzICL1 strain; 4, a ΔGzICL1::acuD strain; 5, a ΔGzMCL1 strain; 6, a ΔGzMCL1::GzMCL1 strain. DNA gel blots in the left, middle, and right panels were hybridized with GzICL1, A. nidulans acuD, and GzMCL1, respectively. The sizes of DNA standards (in kilobases) are indicated on the left.

ICL and MCL activities.

To determine the relative contribution of GzICL1 or GzMCL1 to the enzymatic activities of ICL and MCL, the fungal cell extracts obtained from the ΔGzICL1 and ΔGzMCL1 strains, as well as their wild-type progenitor, Z03643, were examined for ICL or MCL activity (Fig. 2). In both Z03643 and the ΔGzMCL1 strain grown for 24 h in liquid basal medium containing 5% sodium acetate as a sole carbon source, ICL specific activities increased approximately 18 to 20 times over those grown in 2% glucose. In contrast, the ICL activities in both the ΔGzICL1 and ΔGzICL1 ΔGzMCL1 strains grown in acetate were dramatically reduced to levels similar to those observed for growth in glucose (Fig. 2A). Under both growth conditions, the complemented ΔGzICL1::GzICL1 and ΔGzICL1::acuD strains showed ICL activities similar to those in the Z03643 and ΔGzMCL1 strains (Fig. 2A). Similarly, the MCL activities in the Z03643, ΔGzICL1, and ΔGzMCL1::GzMCL1 strains grown in 0.25% propionate were strongly induced (∼10 times) compared with those grown in 2% glucose (Fig. 2B). However, the ΔGzMCL1 and ΔGzICL1 ΔGzMCL1 strains showed approximately 20 times-reduced MCL activities in 0.25% propionate (Fig. 2B). In addition to the gene deletion strains mentioned above, sexual progeny for each gene deletion, which obtained from the sexual crosses described below, showed a similar pattern for the activities for the corresponding enzymes (see Fig. S2 in the supplemental material).

FIG. 2.

Enzyme activities of isocitrate lyase (A) and methylisocitrate lyase (B) in the cell extracts of G. zeae grown in liquid basal medium containing 5% sodium acetate or 0.25% propionate. Fungal strains: WT, the wild-type Z03643 strain; ΔGzICL1, a ΔGzICL1 strain; ΔGzICL1::GzICL1, a complemented ΔGzICL1 strain carrying GzICL1; ΔGzICL1::acuD, a complemented ΔGzICL1 strain carrying A. nidulans acuD; ΔGzMCL1, a ΔGzMCL1 strain; ΔGzMCL1::GzMCL1, a complemented ΔGzMCL1 strain carrying GzMCL1, ΔGzICL1 ΔGzMCL1, a double deletion ΔGzICL1 ΔGzMCL1 strain.

Mycelial growth and other phenotypes of the deletion strains.

To assess the roles of GzICL1 and GzMCL1 in the metabolism of two- or three-carbon compounds or fatty acids, we tested the mycelial growth of the deletion strains on basal medium containing only nonfermentable carbon sources, such as sodium acetate, ethanol, C₃ propionic acid, C₁₂ fatty acid (monolaurate; Tween 20), and C₁₈ fatty acids (Tween 60, linoleic acid, and oleic acid). The ΔGzICL1 strains exhibited about 60% of the wild-type growth on 2% acetate, but they were unable to grow on 5% acetate and other C₂ compounds or fatty acids. However, they showed no significant difference from Z03643 on propionate (Table 1). When an intact copy of GzICL1 was reintroduced into the genome of the ΔGzICL1 strain, its growth defects on acetate and other fatty acids were fully restored; A. nidulans acuD also completely restored the hyphal growth of the ΔGzICL1 strain on the same carbon sources (Table 1). These data suggest that GzICL1 acts as a fungal ICL in the glyoxylate cycle in G. zeae. In contrast, the ΔGzMCL1 strains grew normally on acetate and other fatty acids but failed to grow on propionate, as did the A. nidulans ΔmclA strain lacking MCL (4) (Table 1). However, the ΔGzMCL1 strains showed a different phenotype from the A. nidulans ΔmclA strain when glucose was used as the sole carbon source or propionate was added into glucose medium. The A. nidulans ΔmclA strain grew normally on glucose and was still able to grow on glucose plus propionate (up to 100 mM), although the degree of growth inhibition depended on the propionate concentrations (4). In contrast, the ΔGzMCL1 strains showed significantly reduced growth on glucose alone (Table 1), and their hyphal growth was completely inhibited when even a low amount of propionate (1 mM) was added into glucose. When we reintroduced an intact copy of the GzMCL1 gene back into a ΔGzMCL1 mutant, hyphal growth on propionate was completely restored (Table 1), suggesting that GzMCL1 encodes a putative MCL in the methylcitrate cycle for propionate catabolism. Mycelial growth of the double deletion ΔGzICL1 ΔGzMCL1 strains was abolished on both acetate and propionate (Table 1). All deletion strains showed growth similar to that of the wild type when nutrients were readily available (i.e., PDA and complete medium) (Fig. 3A).

TABLE 1.

Mycelial growth rate^a of the gene deletion strains grown on various carbon sources^b

Carbon sourced	% Residual growth of fungal strainc (±SD)
	WT	ΔGzICL1	ΔGzMCL1	ΔGzICL1 ΔGzMCL1	ΔGzICL1::GzICL1	ΔGzMCL1::GzMCL1	ΔGzICL1::acuD
Glucose (2)	100	100	31 ± 2	29 ± 1	100	100	100
Acetate (2)	100	62 ± 3	96 ± 4	60 ± 2	98 ± 2	96 ± 3	96 ± 4
Acetate (5)	100	0	95 ± 1	0	95 ± 4	98 ± 2	96 ± 4
Tween 60 (2)	100	7 ± 1	99 ± 0	6 ± 2	97 ± 1	97 ± 3	99 ± 1
Linoleic acid (0.25)	100	12 ± 2	96 ± 3	11 ± 2	98 ± 1	97 ± 3	99 ± 1
Propionate (0.25)	100	98 ± 0	0	0	99 ± 1	96 ± 3	98 ± 2
Propionate (0.01)	100	98 ± 2	0	0	98 ± 2	96 ± 2	96 ± 2

^aThe growth rate was determined 7 days after inoculation on each agar plate. All data were obtained from three replicates.

^bThe value obtained with the wild-type strain on each sole carbon source was set to 100%.

^cSee also Fig. 2.

^dNumbers in parentheses indicate the percent concentration of each carbon source.

FIG. 3.

(A and B) Mycelial growth of the wild-type Z03643 strain (upper left of each panel), a ΔGzICL1 strain (upper right), a ΔGzMCL1 strain (bottom right), and a double deletion ΔGzICL1 ΔGzMCL1 strain (bottom left) on PDA (A) or carrot agar (B). (C) Transcript levels of PKS12 in the aerial mycelia of the fungal strains, determined by quantitative RT-PCR.

In addition, all examined deletion strains showed no significant differences from Z03643 for several other phenotypes, including conidiation, conidial germination, and production of mycotoxins, such as deoxynivalenol and zearalenone (data not shown). However, most parts of the aerial mycelia of both the ΔGzICL1 and ΔGzICL1 ΔGzMCL1 strains were much whiter than those of Z03643 on PDA or carrot agar, although normal pigmentation was observed on the agar surface of these deletion strains (Fig. 3A and B); mycelia of Z03643 usually began to produce red pigment 4 to 5 days postinoculation, eventually turning carmine red. The quantitative real-time PCR analysis revealed that the expression of PKS12, which is responsible for the production of the red pigment (aurofusarin) (23), decreased significantly in the ΔGzICL1 and ΔGzICL1 ΔGzMCL1 strains compared to Z03643 (Fig. 3C). The complemented ΔGzICL1 transformants (ΔGzICL1::GzICL1 and ΔGzICL1::acuD) produced red pigment normally, as did Z03643, confirming the effect of ΔGzICL1 on pigmentation of aerial mycelia in G. zeae.

Sexual reproduction of the deletion strains.

To determine whether the gene deletions affected fungal self-fertility, both the ΔGzICL1 and ΔGzMCL1 strains were grown on mating plates (carrot agar). All of the ΔGzICL1 strains examined produced white aerial mycelia during the vegetative growth stage (up to 10 days postinoculation) (Fig. 3B). After the removal of aerial mycelia (i.e., perithecial induction stage), the ΔGzICL1 strains formed either none or few perithecia, indicating that GzICL1 is important for self-fertility in G. zeae. The defect in self-fertility of the ΔGzICL1 strains was completely restored in the ΔGzICL1::GzICL1 and ΔGzICL1::acuD strains. Unlike the ΔGzICL1 strains, the ΔGzMCL1 strains showed no difference from the wild-type strain in both vegetative growth and perithecial formation. The double deletion ΔGzICL1 ΔGzMCL1 strains were very similar to the ΔGzICL1 strains in female fertility and produced few perithecia on carrot agar.

Outcrosses.

To determine the effect of ΔGzICL1 or ΔGzMCL1 on normal meiosis and male fertility in G. zeae, we forced either the ΔGzICL1 (hygromycin B-resistant [HygB^r]) or ΔGzMCL1 (geneticin-resistant [Gen^r]) strain to act as the male parent in outcrosses to a mat1-1 self-sterile G. zeae strain, GzΔMAT1 (HygB and Gen sensitive [HygB^s Gen^s]) on the same mating plates. All of the outcrosses produced similar numbers of mature fertile perithecia as did the outcross of the wild-type strain, indicating that ΔGzICL1 or ΔGzMCL1 had no effect on male fertility in G. zeae. Random ascospores obtained from the outcrosses segregated into parental phenotypes in equal proportions (confirmed statistically by χ² test) for the corresponding drug resistance (HygB^r:HygB^s = 38:38 in the outcross of the ΔGzICL1; Gen^r:Gen^s = 55:45 in the outcross of the ΔGzMCL1). Among progeny carrying both intact MAT idiomorphs from the former outcross, all examined HygB^r progeny, when self-crossed, produced none or few perithecia (<1,000 on a 90- by 15-mm carrot agar plate), as did the ΔGzICL1 parental strain TdGI1-1; all the HygB^s progeny formed >70,000 perithecia on the same plate, as did the wild-type Z03643 strain. This indicates that the significantly reduced female fertility of the progeny was caused by the targeted deletion of GzICL1 from the G. zeae genome. In addition, the HygB^r progeny showed the same phenotype as TdGI1-1 in other characters, such as mycelial growth on various carbon sources and ICL activity. Similarly, all examined Gen^r progeny from the latter outcross showed the same abolished and reduced growth pattern on propionate and glucose, respectively, and the reduced MCL activity as did the parental ΔGzMCL1 strain, TdGI2-3, confirming that ΔGzMCL1 is responsible for the altered phenotypes.

Virulence test.

When single wheat spikelets were point inoculated with a macroconidial suspension of the wild-type Z03643 strain, the head blight symptoms hardly spread into the adjacent spikelets, which was an unexpected result. Instead, we infected both barley and wheat heads with the ΔGzICL1, ΔGzMCL1, or ΔGzICL1 ΔGzMCL1 strains by using the spray inoculation method. When inoculated with the ΔGzICL1 strain TdGI1-1, the disease severity and progression on barley and wheat were similar to those when inoculated with the wild-type Z03643 strain (Fig. 4). Both strains produced typical head blight symptoms that began to appear as early as 3 to 5 days postinoculation and became obvious after 7 days. Similarly, the ΔGzMCL1 strain, TdGI2-3, produced head blight symptoms on both host plants, but the disease severity on barley only seemed somewhat (∼25%) reduced compared to that of Z03643 (Fig. 4). Unlike Z03643 and the single gene deletion ΔGzICL1 or ΔGzMCL1 strains, three independent double deletion ΔGzICL1 ΔGzMCL1 strains showed significantly reduced virulence on barley and wheat. The ΔGzICL1 ΔGzMCL1 strains caused only small necrotic spots on some spikelets of each barley or wheat head even after 10 days following fungal inoculation (Fig. 4). In addition to these transgenic strains, five independent double deletion progeny, which were obtained from an outcross of a ΔGzICL1 ΔGzMCL1 strain to GzΔMAT1, also showed a similar pattern of reduced virulence on both host plants (data not shown).

FIG. 4.

Virulence of the gene deletion stains of G. zeae on barely and wheat heads. (A) Head blight symptoms on barley and wheat heads inoculated with the G. zeae strain Z03643 (WT), a ΔGzICL1 strain, a ΔGzMCL1 strain, and a double deletion ΔGzICL1 ΔGzMCL1 strain. (B) Disease severities caused by the fungal strains, which were determined at 14 days after inoculation by the following formula: ∑[(number of infected spikelets) × (specific symptom scale)]/(total number of spikelets inoculated). Different letters above the bars are significantly different according to the Tukey test (P < 0.05).

Expression patterns of the GzICL1 and GzMCL1 genes.

To determine the transcriptional regulation patterns of the GzICL1 and GzMCL1 genes, transcript accumulations were examined using Northern blot analysis after mycelial growth on various carbon sources (Fig. 5) during the developmental stages for sexual reproduction or pathogenesis (Fig. 6). Significant amounts of GzICL1 transcript were detected on the blot of total RNA extracted from mycelia grown on the two-carbon compounds and the fatty acids examined, but not on glucose. The highest levels of GzICL1 transcript accumulated in mycelia grown on basal medium containing either Tween 20 or 60 (Fig. 5A), whereas GzICL1 expression decreased when the fungal strain was shifted from these conditions to glucose as a sole carbon source (Fig. 5C). In contrast, the expression levels of GzMCL1 were much lower than those of GzICL1 under all of the conditions examined, and gene induction by the carbon sources was not so great (Fig. 5A). However, glucose-induced inactivation of GzMCL1 seemed to occur (Fig. 5C). Some effects of the deletion of one gene on the expression of the other gene were observed. The levels of GzICL1 transcripts decreased significantly when a ΔGzMCL1 strain was grown on the carbon sources, except for propionate, whereas those of the GzMCL1 transcripts were elevated in the ΔGzICL1 strain grown on most carbon sources, except Tween 60 (Fig. 5A). Note that GzICL1 was strongly induced until 4 days after the mycelial plugs of the ΔGzMCL1 strain were inoculated onto basal liquid medium containing propionate, after which time no further mycelial growth of the ΔGzMCL1 strain occurred (Fig. 5B).

FIG. 5.

Expression patterns of GzICL1 and GzMCL1 in fungal mycelia grown in different carbon sources. (A) Total RNA extracted from mycelia of the Z03643 strain (WT), a ΔGzICL1 strain, and a ΔGzMCL1 strain grown in basal liquid medium (MM) and MM containing glucose (Glu), potassium acetate (KOAc), sodium acetate (NaOAc), Tween 20 (T20), Tween 60 (T60), ethanol (EtOH), propionate (PA), or linoleic acid (LA) and separated on an agarose gel, transferred to a membrane, and probed with GzICL1 or GzMCL1. The probes are indicated on the left of each blot. (B) Total RNA gel blots of the fungal strains grown in MM containing 0.25% propionate for 7 days and probed with GzICL1 or GzMCL1. The probes and incubation time (in days) are indicated on the left and above the blots, respectively. (C) Expression patterns of GzICL1 and GzMCL1 in the mycelia of Z03643 grown on basal agar medium (MM) and MM containing glucose (Glu) at 0.25% or 5% glucose or 0.25% Tween 60 (T60) for 3 days, or in those that were previously grown on 0.25% Tween 60 for 3 days and transferred into MM containing 0.25% glucose [T60→Glu (0.25%)] or 5% glucose [T60→Glu (5%)] for an additional 3 days. Ethidium bromide-stained agarose gels (bottom) are shown for comparisons of total RNA loads.

FIG. 6.

(A) Expression patterns of GzICL1 and GzMCL1 in the Z03643 strain grown on carrot agar for sexual reproduction. Refer to Materials and Methods for the detailed procedures for both vegetative growth and perithecial induction. The probes and incubation times (in days) on carrot agar are indicated on the left and above the blot. (B) Gel blot of total RNA from the barley heads infected with Z03643 (WT), a ΔGzICL1 strain, or a ΔGzMCL1 strain and probed with GzICL1 and GzMCL1, respectively. The probes are indicated above the blot. Ethidium bromide-stained agarose gels are shown for comparisons of total RNA loads. (C) Amplification of the transcripts of GzICL1 (left) and GzMCL1 (right) in total RNA used in the RNA gel blot analysis shown in panel B. Days after inoculation are indicated above each blot.

In contrast to gene expression during vegetative growth, GzICL1 expression decreased significantly during the perithecial induction stage on carrot agar (Fig. 6A). Unlike GzICL1, a similar level of GzMCL1 transcripts accumulated during all incubation times, although the transcript level was much lower than that of GzICL1 (Fig. 6A). In barley plants infected with the G. zeae Z03643 strain, the transcription of GzICL1 was strongly induced 6 days after inoculation, but decreased dramatically after day 9 (Fig. 6B). This in planta expression pattern of GzICL1 was also confirmed using RT-PCR analysis. A putative RT-PCR product of GzICL1 was first detected on day 1, increased continuously, and reached the highest level on day 6 (Fig. 6C). The expression of GzICL1 by the ΔGzMCL1 strain in barley decreased similarly to that of vegetative growth (Fig. 6B). The second band of putative GzICL1 RT-PCR products, which was still smaller in size than the genomic DNA band, may indicate alternative splicing of the GzICL1 gene inside the plant, but it needs to be confirmed. Similarly, the in planta expression of GzMCL1 was much lower than that of GzICL1, as in other cases; the expression level was too low to be detected using Northern blot hybridization, but a putative RT-PCR product of GzMCL1 was detected in barley 1 day after inoculation and the level increased continuously until day 9. GzMCL1 expression decreased when the ΔGzICL1 strain was inoculated into barley (Fig. 6C).

DISCUSSION

Several cases for the requirement of ICL genes in fungal virulence led us to investigate the roles of two different ICL orthologs, GzICL1 and GzMCL1, in G. zeae. Sequence similarity suggested that GzICL1 is a typical fungal ICL-encoding gene involved in the glyoxylate cycle for C₂ metabolism, whereas GzMCL1 encodes a putative MCL, rather than an ICL, which is a key enzyme in the methylcitrate cycle for propionate (C₃) metabolism (4, 30). These proposed roles of GzICL1 and GzMCL1 in the two metabolic pathways were genetically confirmed in this study based on the growth phenotypes of the ΔGzICL1 and ΔGzMCL1 strains on C₂ or C₃ compounds as sole carbon sources. The lack of hyphal growth and reduced ICL specific activity in the ΔGzICL1 strain on various C₂ sources demonstrated the effect of ΔGzICL1 in disruption of the glyoxylate cycle, although its growth defect occurred at relatively high concentrations of acetate compared to the other ICL-deleted filamentous fungi (20, 46). The complete restoration of hyphal growth in the ΔGzICL1 strain carrying A. nidulans acuD as well as GzICL1 also support the role of GzICL as an ICL in the G. zeae glyoxylate cycle. The expression pattern of GzICL1 was consistent with those of typical fungal ICL genes. GzICL1 was strongly induced by most nonfermentable carbon sources examined but was repressed when glucose was the sole carbon source. In addition, GzICL1 showed glucose-induced repression, indicating the possibility of catabolite inactivation of the GzICL1 protein, as for other fungal ICLs (1). The absence of growth and reduced MCL activity in the ΔGzMCL1 strains on propionate as the sole carbon source clearly indicates that GzMCL1 encodes a MCL for the utilization of propionate in G. zeae, as for the A. nidulans mclA gene encoding MCL (4). However, the growth phenotypes of the ΔGzMCL1 strains on 2% (111.1 mM) glucose or acetate alone or 2% glucose containing propionate may suggest a different metabolic role for the GzMCL1 protein from A. nidulans mclA. In the A. nidulans ΔmclA strain, growth inhibition occurred only when propionate was added into the glucose or acetate medium, which was explained by the inhibitory effect of methylisocitrate accumulation on the enzymes involved in the tricarboxylic acid (TCA) cycle. This phenotype implies a specific role for MclA in the methylcitrate cycle in A. nidulans (4). In contrast, hyphal growth of the ΔGzMCL1 strain decreased significantly on 2% glucose alone, but not on acetate, indicating that the effect of ΔGzMCL1 on the TCA cycle in G. zeae may be different from that of ΔmclA in A. nidulans. The alterations in the expression pattern of one gene when the other gene is missing may suggest a compensatory network between the GzICL1 and GzMCL1 genes at the transcriptional level. In particular, strong induction of GzICL1 in the ΔGzMCL1 strain on propionate indicates that an interregulatory mechanism between the two genes has been evolutionarily conserved, even though the highly induced GzICL1 expression no longer compensates for the growth defect of the ΔGzMCL1 strain on propionate.

Unlike the effect on fatty acid metabolism, a pronounced cumulative effect of the double deletion was observed for G. zeae virulence on barley and wheat heads, which is clearly different from other fungal pathogens such as C. albicans, L. maculans, and M. grisea (20, 27, 46). The wild-type level of disease development caused by ΔGzICL1 confirms that GzICL1 is not essential for fungal virulence, although its deduced protein sequence is highly similar to other fungal ICLs, which were solely responsible for virulence, and its gene expression is strongly upregulated in barley. The virulence of the ΔGzMCL1 strains on both host plants, similar to that of the wild-type strain, also indicates the functional dispensability of GzMCL1 in fungal virulence, although a slight reduction in disease severity was observed on barley heads (but not on wheat heads). In contrast, the more dramatically reduced disease severity on two hosts inoculated with the double deletion strains clearly indicates that both GzICL1 and GzMCL1 genes (both the glyoxylate and methylcitrate cycles) likely have overlapping or reciprocally redundant functions for full G. zeae virulence in host plants. Because of the difference of gene expression levels, however, it is likely that reciprocity between these two genes is somewhat unbalanced. In addition, the relatively independent contributions of each of these metabolic cycles to the fungal virulence may still exist. The induction of GzICL1 expression in barley 1 day after inoculation (based on RT-PCR) indicates that GzICL1 may be required for the use of C₂ compounds available from plant cell membranes, intracellular stores of lipids, or stored fungal lipids before G. zeae has gained access to nutrients from the host plant tissues (20, 46). The possible function of GzMCL1 in fungal virulence can be postulated based on the role(s) of GzMCL1 discussed above. Possible involvement of GzMCL1 in the TCA cycle may directly affect fungal glucose catabolism for energy acquisition, especially in nutrient-rich plant tissues. In addition, the impairment of propionyl-CoA assimilation throughout the methylcitrate cycle in the ΔGzMCL1 strain would disturb the fungal defense system against propionate or propionyl-CoA in host plants, which may be generated by β-oxidation of odd-chain-length fatty acids in plant lipids. However, it is yet unknown how these two genes (or two metabolic cycles) can redundantly function for only fungal virulence toward host plants. Because they cannot functionally compensate each other for the defects in other phenotypes (e.g., enzyme activity, carbon catabolism, and sexual development), it is unlikely that either of the gene products has a dual role, as ICL1 from Mycobacterium tuberculosis does, as it can clearly function as an MCL as well as an ICL (17).

The functional compensation between GzICL1 and GzICL2 in G. zeae pathogenesis raises a question regarding previous results demonstrating that a single ICL gene is required for fungal virulence. Would the deletion of the MCL or the second hypothetical ICL gene, if present, from the genomes of pathogens with the first ICL gene deleted result in a more severe reduction in or complete abolishment of fungal virulence? The Δicl1 mutants of M. grisea obviously retain the capacity to cause appreciable typical lesions on rice leaves, although they were impaired in functions associated with the prepenetration stage of disease development (46). Therefore, the virulence of M. grisea may be even more severely impaired if the MCL ortholog (MgMCL1; EAA47373) (see Fig. S1 in the supplemental material) is disrupted in the M. grisea Δicl1 mutants. The same strategy could be applied to Cryptococcus neoformans, in which a site-directed mutation in ICL caused no apparent virulence defect in animal models (41), and even to A. fumigatus, which is not dependent on the glyoxylate cycle for invasive aspergillosis (43).

Unlike fatty acid metabolism and fungal virulence, sexual development appears to be affected only by GzICL1 in G. zeae. The formation of fewer perithecia in the ΔGzICL1 strains confirms that only GzICL1 is essential for female fertility in G. zeae. The lack of GzICL1 expression during the later stages of sexual development in the wild-type strain indicates that inactivation of the glyoxylate cycle (or shutdown of an unknown fatty acid catabolic process mediated by GzICL1) may be a prerequisite for switching from the vegetative to the perithecial induction stage in G. zeae. The greater importance of ICL in vegetative growth than in sexual development has been also reported for the brown rot basidiomycete Fomitopsis palustris; ICL activity in mycelia increased markedly at an early stage of vegetative growth but thereafter dropped sharply during the fruiting stage (47). Therefore, the activation of the glyoxylate cycle during the vegetative stage may be necessary for the synthesis of essential cellular constituents that serve as biosynthetic precursors for the formation of perithecia and other sexual structures, as well as for energy supply during the sexual development stage in G. zeae. However, further investigation should examine the importance of isocitrate carbon flow between the glyoxylate and TCA cycles for perithecial induction in G. zeae because the ICL gene is upregulated during fruiting body formation in the ectomycorrhizal ascomycete Tuber borchii (24). A second possibility involves a regulatory role of the metabolic pathway catalyzed by GzICL1 on the biosynthesis of polyketides that may be involved in sexual development. A reduced PKS12 gene transcript level, responsible for the production of the polyketide red pigment, aurofusarin (23), in the white aerial mycelia of the ΔGzICL1 strain, would support this speculation since aurofusarin deficiency leads to elevated production of zearalenone (22), another polyketide whose endogenous levels may affect perithecial induction in G. zeae (Y. T. Kim et al., unpublished data). A third possibility for a role for GzICL1 in sexual development is its involvement in specific metabolism of some fatty acids essential for fungal sexual reproduction. Some specific endogenous fatty acids have long been speculated to be directly involved in fruiting body formation in ascomycetes. In Nectria haematococca, the exogenous supply of linoleic acid enhanced perithecial development, and increased endogenous levels of linoleic acid were detected during perithecial formation (11).

In summary, the glyoxylate cycle mediated by GzICL1 is required for self-fertility, as well as fatty acid metabolism in G. zeae, and plays an important role in fungal virulence toward host plants together with the methylcitrate cycle mediated by GzMCL1 in G. zeae. Our results provide insights into the importance of these two cycles in determining major mycological and pathological traits of G. zeae that were previously unanticipated or unappreciated.