Dissecting Defense-Related and Developmental Transcriptional Responses of Maize during Ustilago maydis Infection and Subsequent Tumor Formation

Abstract

Infection of maize (Zea mays) plants with the smut fungus Ustilago maydis triggers the formation of tumors on aerial parts in which the fungal life cycle is completed. A differential display screen was performed to gain insight into transcriptional changes of the host response. Some of the genes strongly up-regulated in tumors showed a pronounced developmental expression pattern with decreasing transcript levels from basal to apical shoot segments, suggesting that U. maydis has the capacity to extend the undifferentiated state of maize plants. Differentially expressed genes implicated in secondary metabolism were Bx1, involved in biosynthesis of the cyclic hydroxamic acid 2,4-dihydroxy-7-methoxy-2H-1,4-benzoxazin-3-one, and a novel putative sesquiterpene cyclase gene U. maydis induced (Umi)2. Together with the up-regulation of Umi11 encoding a cyclotide-like protein this suggests a nonconventional induction of plant defenses. Explicitly, U. maydis was resistant to 2,4-dihydroxy-7-methoxy-2H-1,4-benzoxazin-3-one but susceptible to its benzoxazolinone derivative 6-methoxy-2-benzoxazolinone. Infection studies of isolated leaves with U. maydis and Colletotrichum graminicola provided evidence for coregulation of Umi2 and PR-1 gene expression, with mRNA levels strongly determined by the extent of fungal colonization within tissue. However, in contrast to Umi2, transcript levels of PR-1 remained low in plants infected with wild-type U. maydis but were 8-fold elevated upon infection with an U. maydis mutant strongly attenuated in pathogenic development. This suggests that U. maydis colonization in planta suppresses a classical defense response. Furthermore, comparative expression analysis uncovered distinct transcriptional programs operating in the host in response to fungal infection and subsequent tumor formation.

The basidiomycete fungus Ustilago maydis is prominent for its ability to induce the formation of large tumors on aerial parts of infected maize (Zea mays) plants, which become filled with masses of black-pigmented sexual teliospores (Banuett and Herskowitz, 1996; Kahmann et al., 2000). U. maydis is a member of the smut fungi, which infect a large number of dicotyledonous and monocotyledonous plants and cause severe economic losses to growers (Banuett and Herskowitz, 1996; Martínez-Espinoza et al., 2002). Within this group, U. maydis represents the most tractable system offering a comprehensive repertoire of molecular tools (Kahmann and Kämper, 2004). For reasons unknown, U. maydis infection is confined to young aerial tissue parts of maize (Wenzler and Meins, 1987, and refs. therein). Initial growth proceeds primarily intracellularly through epidermal cells, and the fungus is surrounded by the intact host cell plasma membrane at this stage (Snetselaar and Mims, 1992, 1994). Early disease symptoms of infected maize plants are chlorosis (yellowing of tissue), the formation of anthocyanin, and stunting while signs of defense responses are not detected at the ultrastructural level (Snetselaar and Mims, 1993; Banuett and Herskowitz, 1996). Tumor formation is associated with plant cell enlargement and proliferation (Callow and Ling, 1973; Snetselaar and Mims, 1994). Within the tumor fungal cells are embedded in parenchymatous, thin-walled cells, which lack plastids (Callow and Ling, 1973).

The pathogenic form of U. maydis is the filamentous dikaryon, which is generated after fusion of two compatible, haploid sporidia on the leaf surface. The dikaryotic hyphae is able to develop an appressorium-like swelling at the tip, which allows penetration (Snetselaar and Mims, 1993). Subsequent fungal differentiation is intimately coupled to growth within host tissue and includes extensive branching, karyogamy, hyphal fragmentation, and teliospore formation (Banuett and Herskowitz, 1996; Kahmann et al., 2000). The sexual life cycle of U. maydis is governed by a tetrapolar mating system consisting of the a and b mating type loci (Kronstad and Staben, 1997; Kahmann et al., 2000). The biallelic a locus encodes a pheromone/receptor system required for sporidial fusion (Bölker et al., 1992; Spellig et al., 1994). The multiallelic b locus encodes the bE/bW homeodomain-containing proteins, which in nonallelic combinations dimerize to an active transcription factor complex crucial for pathogenic development (Gillissen et al., 1992, and refs. therein; Kämper et al., 1995).

While recent research has provided detailed insight into the complexity of signaling pathways controlling morphogenesis and pathogenicity of U. maydis (Basse and Steinberg, 2004; Kahmann and Kämper, 2004), the molecular events operating on the host side in response to fungal infection are unknown. Interesting questions concern the molecular events coinciding with tumor development as well as the induction of genes related to defense based on the apparent ability of U. maydis to escape detection by the host. To address the interplay between U. maydis and its host, differential display analysis was performed comparing expression in leaf tumor and noninfected leaf tissue (Basse et al., 2000). The fungal genes retrieved from this approach were mig1 and mig2-1 to mig2-5, which display a pronounced plant-specific expression profile and encode small, secreted, Cys-containing proteins of unknown function, as well as udh1, which encodes a functional steroid 5α-reductase (Basse et al., 2000, 2002a, 2002b; Farfsing et al., 2005). Here, I report the identification of maize genes that exhibit a marked differential expression profile during U. maydis infection. This study provides molecular evidence that U. maydis has the capacity to extend the undifferentiated state of infected tissue and yields insight into putative defense responses as well as distinct regulatory programs triggered during U. maydis infection.

RESULTS

Isolation of U. maydis Induced and U. maydis Repressed Genes

A differential display approach was performed comparing transcripts in 6-d-old leaf tumor and noninfected leaf tissue of maize plants (Basse et al., 2000). At this stage, tumor development is in progress. Of more than 7,000 cDNA fragments amplified, 18 displayed a differential pattern as verified by Southern-blot hybridization (data not shown; see “Materials and Methods”). Differential expression of these fragments was verified by RNA blot or reverse transcription (RT)-PCR analysis. This revealed seven maize genes >20-fold up-regulated, three maize genes <10-fold up-regulated, and two maize genes markedly down-regulated in leaf tumor tissue compared to control tissue of the same age (Table I). Identified genes were designated U. maydis induced (Umi) and U. maydis repressed (Umr), respectively. The majority of fragments identified by differential display corresponded to known maize genes (Table I) with functions related to primary metabolism (Umi9, Umi13, and Umr2), the formation of the antimicrobial compound 2,4-di-hydroxy-7-methoxy-2H-1,4-benzoxazin-3-one (DIMBOA; Umi7/Bx1), and development (Umi4 and Umi6). The novel Umi5 and Umi8 cDNA sequences (GenBank accession nos. AY679127 and AY679128) encode proteins with putative functions as pyruvate kinase and CTP synthase, respectively, and the product encoded by Umi12 (accession no. AY679130) is highly similar to calmodulins. The Umi2 and Umi11 cDNA fragments contained nontranslated 3′ regions that provided no clues to their functions. Therefore, full-length cDNA clones were isolated from these genes (see “Materials and Methods”). The Umi11 sequence (accession no. AY679129) contained an open reading frame (ORF) encoding a protein of 90 amino acids. Protein sequence analysis using the protein families database (PFAM) assigned the region from positions 56 to 88 to the cyclotide domain of cyclotide family proteins (Fig. 6; “Discussion”). The deduced amino acid sequence of Umi2 (accession no. AY647253) displays strongest similarities to the recently identified sesquiterpene cyclases TPS4 and TPS5 from maize (58% and 59% identity, respectively) and corresponds to the maize sesquiterpene cyclase TPS6 (J. Degenhardt, personal communication; Table I).

Table I.

Differentially expressed maize genes in U. maydis-infected tissue

Gene	x-Fold Activationa	Isolated cDNA and ORF Regionb	BLASTX Annotation	BLASTX Score	GenBank Accession No.
Umi2 (tps6)	33	1,742 (96–1,742)	Maize terpene synthase TPS5 (AAS88575)	1e-180	AY647253
Umi4	2.5	317 (1–93)	Maize ribosomal protein L17	Identical	AF034948
Umi5	>100	636 (1–441)	Oryza sativa putative pyruvate kinase (NP_917361)	5e-70	AY679127
Umi6	60	364 (1–52)	Maize MFS18	Identical	X67324
Umi7	21	228	Maize Bx1	Identical	X76713
Umi8	>100	190 (1–190)	O. sativa putative CTP synthase (NP_917309)	2e-18	AY679128
Umi9	50	382 (1–382)	Maize Lys ketoglutarate reductase	Identical	AF003551
Umi11	>100	542 (34–306)	Viola odorata cycloviolacin O10 (P58442)	1.5	AY679129
Umi12	5.0	339 (1–312)	O. sativa putative calmodulin (AAP13012)	1e-49	AY679130
Umi13	3.1c	264	Maize adenine phosphoribosyl transferase APT1	Identical	AY485263
Umr1	0.35	366 (1–143)	Maize β-d-glucosidase (glu2)	Identical	ZMU44087
Umr2	0.13	322 (1–162)	Maize P700 chlorophyll a-protein PSI A2	Identical	M11203

^aDetermined by RNA-blot analysis (see Fig. 1) comparing transcript levels in tumor tissue 6 dpi and noninfected control tissue of the same age. The ratios for Umr1 and Umi4 transcripts were determined 4 and 8 dpi, respectively (strains FB1/FB2, data not shown).

^bLengths of isolated cDNA fragments are indicated in base pairs. ORF-spanning regions are indicated in parentheses.

^cDetermined by RT-PCR analysis comparing transcript levels in tumor tissue 6 dpi (strains FB1/FB2, data not shown) and noninfected control tissue of the same age.

Figure 6.

Disulfide bridge arrangement of cyclotide domains and sequence comparison with UMI11. The putative cyclotide domain deduced from the Umi11 ORF (amino acid positions 56–88) was aligned with those of the mature Kalata B5 (subfamily 1), cycloviolacin O10 (subfamily 1), and cycloviolacin O12 (subfamily 2) proteins. The six highly conserved Cys residues are shaded black. Lines (top) refer to internal disulfide bridges. The extended line (bottom) refers to the ring closure between Ile and Pro. Identical or highly similar (Ile/Leu, Ser/Thr, Arg/Lys) positions dominating a column are shaded gray. Gaps were introduced to maximize the alignment.

Expression Kinetics of Umi and Umr Genes

To examine temporal expression of identified genes by RNA-blot analysis RNA was isolated in 2-d intervals from infected and noninfected leaf tissues of the same age. Marked transcript levels of Umi5, 6, 8, 11, and 12 were already observed in noninfected 6-d-old maize seedlings (day 0 in Fig. 1) and levels successively declined to non- or weakly detectable. In contrast, transcript levels of these genes raised again in infected samples. While transcripts of Umi2, 5, 6, and 8 reached maximum levels between 4 and 6 d after inoculation (dpi), those of Umi7, 9, and 11 peaked between 6 and 8 dpi. Transcript levels of Umr2 steadily declined after 2 dpi in infected compared to noninfected tissue. The decrease in transcript levels of the down-regulated gene Umr1 was less pronounced and confined to 4 dpi. As expected, none of the Umi gene fragments hybridized with fungal control RNA isolated from dikaryotic hyphae (Fig. 1).

Figure 1.

Kinetics of Umi/Umr gene expression. A, RNA-blot analysis of various maize genes identified by differential display analysis. Maize seedlings were inoculated with either FB1 or a mixture of FB1/FB2 sporidia. RNA was isolated from tissue immediately after inoculation (0) or after 2, 4, 6, and 8 d (numbers on top). Tissue from mock-infected plants and infected plants prior to tumor formation was harvested from leaf blades below the inoculation site. Lane c, Control RNA from dikaryotic hyphae (strain mixture FB1/FB2). Each gel was loaded with the same RNA preparations. B, RNA-blot analysis for Umi12. Identical amounts of the same samples (samples 8 dpi were omitted) as performed in A were loaded. A and B, Staining with methylene blue reflects the amounts of total RNA loaded (see Table I for quantification of signals).

Developmental Regulation of Umi Genes

The relatively high transcript levels of Umi genes in noninfected 6-d-old maize seedlings and their temporal decline (Fig. 1) suggested developmental regulation. To verify this assumption, noninfected plants of various ages were dissected into three to four shoot segments including leaf blade tissue to compare expression of the Umi2, 6, 7, 8, and 11 genes with expression in leaf blade tumor tissue 6 dpi. These genes markedly differed in relative expression signals in noninfected tissue (Fig. 1). Transcript levels of Umi6, 8, and 11 strongly declined with increasing distance from the shoot base and were lowest in leaf blade tissue irrespective of the seedling age (Fig. 2A, lanes 3, 6, and 10). This decrease was most pronounced for Umi6. In contrast to Umi8 and 11, whose transcript levels were highest in leaf blade tumors (lane 11), quantification of the Umi6 signals revealed a 53-fold increase in tumor tissue relative to noninfected blade tissue of the same age (lanes 11 and 10) but a 2,000-fold increase in basal shoot tissue (lanes 7 and 10). Transcripts of the maize actin gene MAc1, which served as control, were only weakly elevated in young shoot or tumor tissue (Fig. 2B). For Umi7, a developmental influence on gene expression was essentially confined to 5-d-old seedlings, whereas Umi2 transcripts were relatively weak in noninfected tissues and revealed no decrease from basal to apical segments (Fig. 2B). Maximum levels of Umi2 and Umi7 transcripts were detected in leaf blade tumor tissue (lane 11). Collectively, this shows that identified Umi genes were differently regulated during development and suggests that U. maydis alters the developmental program in infected plants.

Figure 2.

Developmental influence on Umi gene expression. A, Expression analysis of Umi6, 8, and 11. B, Expression analysis of Umi2, 7, and MAc1. A and B, Samples were taken from 5-d-old (1–3), 9-d-old (4–6), and 12-d-old seedlings (7–11) for isolation of total RNA. The basal-most 2 cm of the shoot from which the coleoptile leaf had been removed (1, 4, 7), 2 to 4 cm from the shoot base (2, 5, 8), 4 to 7 cm from the shoot base (9), leaf blade tissue of the first and second leaves (3), leaf blade tissue of the second and third leaves (6, 10), leaf blade tumor tissue of the second and third leaves 6 dpi (strains FB1/FB2; 11). Bars reflect relative expression levels individually calculated for each gene from the signals obtained by RNA-blot analysis. The sum of all bars assigned to a single gene corresponds to 100% (log scale). Umi11 and 2 (white bars, left), Umi8 and 7 (gray bars, middle), Umi6 and MAc1 (black bars, right).

Expression of Umi Genes in Dependence on Tumor Formation and Fungal Colonization

The finding that induction of Umi gene expression in infected tissue coincided with tumor formation (Fig. 1) raised the question as to whether this was prerequisite for their induction. For this reason expression was compared in different parts of infected plants including tumor and nontumor tissues. RNA was isolated (6 dpi) from leaf blade tumors (T1) and bordering green nontumor tissue without signs of infection (T2), chlorotic nontumor leaf blade tissue below the inoculation site (T3), leaf blade tissue lacking disease symptoms distal from tumors (T4), and leaf blade tissue of noninfected plants (T5). As judged from RNA-blot analysis (Fig. 3A), expression of Umi2 as well as of Umi7 was induced to similar levels in tumor (T1) and chlorotic nontumor tissue (T3). Weaker Umi7 transcript levels were detected adjacent to the tumor (T2) as well as at distal sites (T4), while Umi2 transcripts were not detected in these samples by RNA-blot analysis (Fig. 3A). In contrast, relative to the strong Umi6, 8, and 11 transcript levels in tumor tissue (T1), only faint Umi8 levels were detected in samples T2 to T4 and those of Umi6 and 11 were absent or hardly detectable (Fig. 3A).

Figure 3.

Expression of Umi genes related to fungal colonization and tumor induction. Maize seedlings were inoculated either with mixtures of strains FB1/FB2 (wild type, T1–T4), FB1Δmrb1/FB2Δmrb1 (Δmrb1, T6, and T7), AB311/AB312 (T8 and T9), or strain FB1 (T10). All samples were isolated from leaf blade tissue of the third and fourth leaves either 6 dpi or from untreated control plants of the same age (T5). T1, Leaf tumor tissue; T2, green leaf tissue lateral to tumors showing no signs of infection; T3, T7, and T9, tissue 2 to 5 cm below the inoculation site; T4, green leaf tissue distal (>2 cm) from tumors; T6, T8, and T10, tissue around (<1 cm) the inoculation site. A, RNA-blot analysis of Umi2, 6, 7, 8, 11, and PR-1. Staining with methylene blue reflects the amounts of total RNA loaded. B, RT-PCR analysis of U. maydis genes ip, mig2-5 (white triangles) and of maize genes Umi2 and H2B (control for integrity of RNA in all reactions). Number of cycles performed: 28 for mig2-5, Umi2, and H2B; 32 for ip.

To examine the presence of Umi2 transcripts in samples T2 and T4 and to correlate the expression of Umi2 and Umi7 with fungal amounts in infected tissues, semiquantitative RT-PCR analysis was performed using primers against Umi2 and the U. maydis genes ip and mig2-5, respectively. The ip gene (Broomfield and Hargreaves, 1992) is constitutively expressed and allows quantification of the overall presence of the fungus. Expression of mig2-5, which is strictly linked to biotrophic growth (Basse et al., 2002a), provides for a suitable marker for the fungal presence within plant tissue. Transcript levels of mig2-5 were comparable in samples T1 and T3 were reduced in sample T2 and lowest in T4 (Fig. 3B). Umi2 transcripts were also detected by RT-PCR analysis in the weakly infected sample T2 but were apparently absent in sample T4. As expected, no Umi2 signal was detected in untreated control tissue (T5). Together, this indicated that expression of both genes Umi7 and Umi2 was determined by the extent of fungal colonization.

Expression of Umi Genes in Response to Infection with an U. maydis Proliferation Mutant and in Dependence on Fungal Penetration

Infection by U. maydis is a multistep process including the formation of intimate infection structures like appressoria and penetration pegs on the leaf surface followed by hyphal proliferation and differentiation in planta. The availability of U. maydis mutants defective at different developmental stages should allow the further dissection of prerequisites for Umi gene induction. U. maydis Δmrb1 mutant strains have a severe proliferation defect in planta and cause only attenuated and rare induction of tumors (Bortfeld et al., 2004). To further relate Umi2 and Umi7 gene expression to fungal proliferation, plants were inoculated with Δmrb1 mutant strains and tissue was isolated around the inoculation site (T6) and from chlorotic areas below the inoculation site (T7) 6 dpi. Unexpectedly, RNA-blot analysis revealed that the Umi2 signal in sample T7 was at least as strong as in corresponding tissue infected by wild-type strains (T3), whereas a Umi7 signal was not detected in this sample. As expected, expression of the Umi6, 8, and 11 genes was not detected in response to infection with Δmrb1 mutants (Fig. 3A). The strong Umi2 expression in sample T7 was confirmed by RT-PCR analysis, which additionally revealed a weak mig2-5 signal relative to sample T3 indicative for low amounts of fungal hyphae within plant tissue (Fig. 3B). Moreover, a marked Umi2 signal was also detected in tissue around the inoculation site (T6), in which only few penetration events are expected, consistent with hardly detectable mig2-5 and strong ip signals (Fig. 3B).

The sensitive induction of Umi2 expression after U. maydis infection raised the question as to whether this required fungal penetration. For this reason plant infections were performed with the compatible strains AB311 and AB312. In these strains, kpp6, which encodes a mitogen-activated protein kinase essential for penetration, is replaced by a variant allele encoding a nonactivatable form of the protein. Resulting dikaryotic hyphae develop appressoria, with the majority producing only short filaments that fail to enter the plant (Brachmann et al., 2003). RT-PCR analysis revealed high levels of ip transcripts in plant tissue either around (T8) or below (T9) the inoculation site, whereas only a faint mig2-5 signal was detected below the inoculation site (Fig. 3B). Remarkably, a weak Umi2 signal was detected in tissue around, but not below, the inoculation site (Fig. 3B), revealing that Umi2 expression was weakly triggered from the plant surface. To exclude that this was caused by haploid sporidia on the leaf surface, plants were inoculated with strain FB1. Its surface presence was reflected by a strong ip signal and the absence of the mig2-5 signal (T10 in Fig. 3B). This showed that sporidia on the leaf surface were unable to trigger Umi2 expression. As expected, expression of remaining Umi genes was not induced by hyphal surface contacts as judged from RNA-blot analysis (Fig. 3A).

Comparison of Umi2 with PR-1 Gene Expression

Prompted by the strong increase in Umi2 expression in response to fungal colonization, expression of the defense-related maize PR-1 gene was investigated in plants infected with U. maydis wild-type and mutant strains. This gene is strongly expressed during a resistance response in incompatible interactions, while in successful infections expression is weak and attenuated (Morris et al., 1998). PR-1 gene expression was examined in the tissue probes T1 to T10 (Fig. 3A). RNA-blot analysis showed relatively weak transcript levels in samples T1 and T3 representing wild-type infected tumor and leaf parts, respectively. Remarkably, however, strong PR-1 transcript levels were detected in leaf sections infected with Δmrb1 mutant strains (T7). In contrast, PR-1 expression was hardly detectable when plants were inoculated with the AB311/312 strain combination. This indicated that PR-1 gene expression displayed a similar pattern as seen for Umi2; however, while quantification of PR-1 signals in sample T7 revealed an 8-fold increase over transcript levels in the corresponding sample T3, this difference was only 1.6-fold for the Umi2 transcript.

Investigation of Defense-Related Gene Expression in Isolated Maize Leaves Challenged Either with U. maydis or Colletotrichum graminicola

To further compare Umi2 with PR-1 gene expression in dependence on fungal colonization and to examine whether activation of defense-related Umi genes is specific for the U. maydis interaction, detached maize leaves were treated with the U. maydis strain JF1 as well as with the hemibiotrophic fungus Colletotrichum graminicola (see “Materials and Methods”). Strain JF1 is solopathogenic and thus is not relying on a mating partner to become pathogenic. In addition, based on the presence of an enhanced green fluorescent protein (eGFP) reporter construct under control of the mig2-5 promoter, hyphal growth within plant tissue can be assessed from fluorescence microscopy (Farfsing et al., 2005). This revealed that from 3 d after spotting of JF1 sporidia on the leaf surface, rare hyphal growth was observed in epidermal and underlying cells (Fig. 4A; data not shown). In contrast, Chlorazol Black E (CBE) staining showed intensive fungal colonization 2 d after spotting of C. graminicola conidia, and 1 d thereafter hyphae had extensively ramified throughout the entire leaf (Fig. 4A; data not shown). The observed infection progress of C. graminicola was coincident with a previously reported infection study (Thon et al., 2002). Consistently, RNA-blot analysis revealed only weakly elevated Umi2 transcript levels, which were confined to the 5-d time point, in tissue infected with strain JF1 (Fig. 4B), whereas in tissue challenged with C. graminicola Umi2 levels were strongly increased at 2 and 3 d and subsequently declined (Fig. 4B). Furthermore, marked Umi2 transcript levels were already detected 1 d after application of C. graminicola, when the leaf surface was covered with appressorial structures prior to colonization within tissue (Fig. 4, A and B). Remarkably, the expression pattern of Umi2 perfectly coincided with that of PR-1 in leaves infected with both U. maydis and C. graminicola (Fig. 4B), providing evidence for common regulation. By contrast, Umi7 or Umi11 transcript levels were induced neither in response to U. maydis infection nor during extensive C. graminicola colonization in isolated leaves. The increased Umi7 transcript levels seen at the 1-d time point were also detected in untreated samples and might be a stress response to leaf isolation (Fig. 4B). Although strain JF1 has the potential to trigger tumor formation in infected plants, this is not observed in infections of detached leaves and may provide an explanation for the absence of Umi11 induction at the 5 dpi time point. The absence of Umi7 induction despite colonization either by strain JF1 or C. graminicola may indicate that a developmental component as triggered during the U. maydis plant infection participates in Umi7 regulation.

Figure 4.

Defense-related gene expression in response to U. maydis and C. graminicola infection. A, Hyphal growth within plant tissue and C. graminicola penetration structures. Pictures were taken from U. maydis infected tissue 5 d after application of sporidia (JF1) and from C. graminicola infected tissue 1 d (bottom) and 3 d (top) after application of conidia (Cg). Maize tissue samples were assayed by epifluorescence (GFP; exposure time = 2 s) and/or differential interference contrast light microscopy (DIC). C. graminicola infection structures were stained with CBE. Hyphae and appressoria are marked by arrowheads. Bars = 10 μm. B, Time course of Umi and PR-1 expression during fungal infection. For each time point, RNA was isolated from three leaves either untreated or identically treated with U. maydis strain JF1 (JF1) and C. graminicola (Cg), respectively. Numbers on top refer to days after fungal application. Staining with methylene blue reflects the amounts of total RNA loaded.

Increased Formation of DIMBOA in Tumors of Infected Plants and Investigation of Antifungal Activity of Cyclic Hydroxamic Acids

To address the question of whether the induction of Umi7 (Bx1) expression is linked to DIMBOA production, plants were inoculated either with the FB1/FB2 strain combination or with strain FB1 alone. Eight independent samples each from leaf blade tumors or mock-infected blade tissue collected 6 dpi were extracted for HPLC analysis (see “Materials and Methods”). This revealed a DIMBOA concentration of 1.81 ± 0.38 nmol/mg tissue in tumors compared to a concentration of 0.54 ± 0.20 nmol/mg tissue in mock-infected controls. Based on the resulting DIMBOA concentration of 0.37 mg/mL in tumor tissue, DIMBOA was tested in various concentrations on growth inhibition of U. maydis on solid medium (Fig. 5). In addition, 6-methoxy-2-benzoxazolinone (MBOA), which emerges from spontaneous degradation of DIMBOA and is also an effective antimicrobial compound (Glenn et al., 2002, and refs. therein), was included in this assay. Excitingly, this revealed that while U. maydis strains FB1 and FB2 fully resisted DIMBOA at concentrations ≤1 mg/mL, U. maydis was susceptible to MBOA; colony growth of both strains was strongly reduced at a concentration of 0.3 mg/mL and blocked at a concentration of 1 mg/mL (Fig. 5). Furthermore, DIMBOA, but not MBOA-containing medium, turned reddish after U. maydis cultivation (data not shown), suggesting that DIMBOA is enzymatically degraded by U. maydis.

Figure 5.

Effect of cyclic hydroxamic acids on colony growth of U. maydis. Growth of U. maydis strains FB1 and FB2 on solid PD medium amended with various concentrations of DIMBOA (D), MBOA (M), and Trp (T) after 24 h. The final concentration of these compounds is indicated on the left. Trp was included as control indole compound in this assay.

DISCUSSION

While U. maydis has developed a prime model organism to unravel molecular mechanisms of fungal phytopathogenicity (Kahmann and Kämper, 2004), the host side has not been addressed in molecular terms. A differential display analysis was performed to gain insight into how maize plants respond to U. maydis infection at the transcriptional level. This revealed drastic changes in transcript levels of genes related to metabolism and development and uncovered putative defense responses of the host plant.

Umi Genes Related to Primary Metabolism and Development

The pronounced decline of Umr2 transcripts encoding a central component of PSI is in agreement with the disappearance of plastids during tumor formation in U. maydis-infected plants (Callow and Ling, 1973). This may be related to a switch in the metabolic program, which is also indicated by the strong up-regulation of the Umi5 and Umi8 genes predicted to encode pyruvate kinase and CTP synthase, respectively. Umi9 encodes the bifunctional enzyme Lys-ketoglutarate reductase/saccharopine dehydrogenase (LKR/SDH) initiating the α-aminoadipic acid pathway of Lys catabolism. Abundant transcripts of the corresponding gene ZLKRSDH were shown to be confined to the endosperm, and it was discussed that regulation of ZLKRSDH was subject of a transcriptional control mechanism related to amino acid metabolism and synthesis of storage proteins (Kemper et al., 1999). Based on the strong ZLKRSDH expression in both endosperm and tumor tissue, it will be interesting to investigate common transcriptional processes.

Induction of the Umi4 and Umi6 genes may be related to developmental processes. Umi4 corresponds to the ribosomal protein L17 of maize, which is 86% identical to barley (Hordeum vulgare) L17.2. The L17.2 gene is predominantly expressed in regions of meristematic and young, rapidly growing cells in both leaf and root (Madsen et al., 1991). Umi6 corresponds to MFS18, another developmentally regulated gene that is drastically up-regulated in developing maize tassel tissue (Wright et al., 1993). MFS18 encodes a small, putative structural cell wall-associated protein with 66% identity to the barley 8s protein sequence. The 8s gene is expressed in leaf sections directly adjacent to the basal meristem, while expression is absent in the expanded leaf blade (Schünmann and Ougham, 1996). This is consistent with the finding that Umi6 expression is highest in the most basal shoot tissue. For the 8s gene, a role in the early stages of elongation growth was suggested (Schünmann and Ougham, 1996). The basal shoot segment is enriched in dividing cells, while upper segments are mainly composed of expanding leaf cells and mature leaf tissue is fully expanded and differentiated (Smith et al., 2001). The observation that expression of the Umi6, 8, and 11 genes is confined to basal shoot tissue in noninfected seedlings, whereas these genes are strongly expressed in tumors present in expanded leaf blades, indicates that U. maydis has the capacity to interfere with differentiation of infected tissue. Previous investigations have shown that U. maydis-induced tumor formation is essentially confined to immature expanding tissue at the leaf base (Wenzler and Meins, 1987). Tissue at the base of developing leaves is composed of dividing and expanding cells (Sylvester et al., 1990; Smith et al., 2001). Although extending this stage would be sufficient to maintain expression of developmentally regulated Umi genes, the possibility remains that U. maydis triggers dedifferentiation of more mature tissue.

Umi Genes Related to Secondary Metabolism and Defense

The strongly elevated Bx1 transcript levels and increased amounts of the cyclic hydroxamic acid DIMBOA in tumor tissue relative to noninfected tissue indicate the induction of a plant defense response. Bx1 catalyzes the conversion of indole-3-glycerol phosphate to indole, representing the initial step in the formation of DIMBOA (Frey et al., 1997). Increased amounts of DIMBOA in young tissue have been reported, and DIMBOA levels correlate with resistance to a number of fungal pathogens (Niemeyer, 1988; Sicker et al., 2000; Glenn et al., 2002, and refs. therein). In particular, antimicrobial activity of DIMBOA and benzoxazolinone derivatives MBOA and 2-benzoxazolinone (BOA) against Fusarium species pathogenic on maize has been reported (Glenn et al., 2002, and refs. therein). While resistance of Fusarium verticillioides to BOA depended on its ability to detoxify this compound, growth of sensitive strains was blocked at a BOA concentration of 1 mg/mL and strongly reduced at a concentration of 0.5 mg/mL (Glenn et al., 2002). Intriguingly, U. maydis fully resisted DIMBOA; however, it was severely affected by its benzoxazolinone derivative at concentrations ≥0.3 mg/mL, and thus U. maydis displayed a similar dose response to MBOA as F. verticillioides to BOA. Resistance of U. maydis to DIMBOA may be explained by the ability of detoxification. This now provides a means by which to isolate sensitive mutants and uncover responsible genes.

Umi2, which has been independently identified as tps6 (J. Degenhardt, personal communication), encodes a putative sesquiterpene cyclase that displays strongest similarities to maize TPS4 and TPS5. These enzymes catalyze the biosynthesis of a broad spectrum of volatile sesquiterpenes from farnesyl diphosphate (Köllner et al., 2004). Since sesquiterpenes are also related to plant defense against microorganisms (Bell, 1986; Threlfall and Whitehead, 1991; Bohlmann et al., 1998), it will be interesting to determine the spectrum of compounds produced by UMI2/TPS6 as well as possible effects of these compounds on U. maydis growth.

The presence of a putative cyclotide domain in the amino acid sequence deduced from Umi11 points to the existence of plant cyclotide family proteins in monocotyledonous plants. Genes encoding cyclotides have not been found in the genome of Arabidopsis (Arabidopsis thaliana) and are presently known only from the Rubiaceae and Violaceae families (Trabi and Craik, 2004). Cyclotides are divided in two subfamilies according to amino acid variations between conserved Cys residues (Craik et al., 1999; Trabi and Craik, 2004). The predicted maize cyclotide contained sequence elements of both subfamilies (Fig. 6), suggesting a common ancestor protein. Mature cyclotides released from the secretory pathway are cyclic head-to-tail proteins comprising 29 to 31 amino acids and include six highly conserved Cys residues that form a cystine knot (Fig. 6; Craik et al., 1999; Jennings et al., 2001). A secretory signal sequence comprising the N-terminal 30 amino acids is also predicted for UMI11 according to program TargetP (data not shown). Reminiscent of Umi11, expression of the cyclotide encoding gene B1 was enhanced in young shoot tissue of Oldenlandia affinis compared to mature leaves (Jennings et al., 2001), although the difference was not as pronounced as what has been observed for Umi11. Cyclotides are structurally related to plant defensins, which occur ubiquitously throughout the plant kingdom (Craik et al., 1999, and refs. therein; Trabi and Craik; 2004), and antimicrobial activity against several fungal species has been shown (Tam et al., 1999; Jennings et al., 2001). Thus, it will be interesting to determine whether this also accounts for the putative maize cyclotide.

Finally, the significant up-regulation of Umi12, whose deduced amino acid sequence shows strong similarities to calmodulins, points to enhanced Ca²⁺ signaling after U. maydis infection. Whether this is related to plant defense (Martin, 1999) or development (Yang and Poovaiah, 2003) remains to be shown.

Although this study implies induced plant defenses, it is likely that U. maydis has developed mechanisms to cope with them based on its close association with its host (Snetselaar and Mims, 1992, 1994). Such growth, which does not disrupt plant cell integrity, would prevent the release of cyclic hydroxamic acids (Glenn et al., 2002, and refs. therein), and therefore these compounds might be directed against necrotrophic microbes causing cellular damage. The induction of the defense responses described is not linked to strong PR-1 gene expression. This is consistent with the observation that the coordinately regulated maize PR-1 and PR-5 genes are more rapidly and strongly induced in incompatible than in compatible interactions with the rust fungus Puccinia sorghi (Morris et al., 1998). The finding that U. maydis Δmrb1 mutants triggered pronounced PR-1 expression despite low fungal colonization relative to wild-type infection leads to the assumption that U. maydis escapes induction of a classical host defense due to its proliferation capacity.

Dissecting Transcriptional Responses to U. maydis Infection

The investigation of Umi gene expression in relation to the extent of fungal colonization and tumor formation gave insight into distinct programs operating during U. maydis infection (Fig. 7). In plants infected with wild-type strains, expression of Umi2 and Umi7 was already induced during hyphal growth in nontumor tissue and transcript levels of both genes were influenced by the extent of fungal colonization. However, these genes are differently regulated as documented by the absence of a Umi7 signal in tissue infected with Δmrb1 mutant strains as well as in isolated leaves infected either with U. maydis or C. graminicola. This may indicate that induction of Umi7 expression is not merely a consequence of fungal colonization but includes a developmental component specifically triggered during the U. maydis-host interaction. By contrast, induction of Umi2 expression coincided with fungal colonization and was not specific for U. maydis. Furthermore, as inferred from infections with C. graminicola, Umi2 expression was markedly induced prior to colonization within plant tissue. This early induction may be explained by intimate contacts resulting from efficient formation of appressorial structures on the leaf surface. This conclusion is consistent with the finding of weakly induced Umi2 transcript levels triggered by the strain combination AB311/AB312 (see Fig. 3B), which sticks in penetration attempts on the leaf surface (Brachmann et al., 2003). Intriguingly, expression analysis of infected leaves pointed to common mechanisms involved in Umi2 and PR-1 gene regulation. On these grounds it will be exciting to explore the strategy U. maydis adopts to specifically suppress PR-1 expression during the host interaction.

Figure 7.

Proposed scheme for distinct signaling pathways during U. maydis infection. Expression of Umi2 is strongly induced during hyphal colonization within plant tissue (arrows) and weakly by intimate contacts on the surface (dashed arrow). Umi2 and PR-1 are at least in part coregulated and not specific for U. maydis infection. Expression of Umi7 is influenced by hyphal growth within plant tissue and developmental signals as triggered during U. maydis colonization. Induced expression of Umi6, 8, and 11 results from developmental responses to U. maydis infection. White arrow heads refer to defense-related functions, while remaining Umi genes might have functions in fungal support. See text for further details.

Increased expression of Umi6, 8, and 11 was a consequence of tumor formation and thus a secondary response to U. maydis infection. Differences in their expression patterns in basal shoot relative to tumor tissue point to additional signals specifically operating in tumors and influencing expression of developmentally regulated genes. It will be a challenge to uncover the underlying programs and to investigate how these relate to the morphological transition caused by the U. maydis interaction.

MATERIALS AND METHODS

Plant Material, Strains, and Growth Conditions

Maize (Zea mays) infections were done with the variety Early Golden Bantam (Olds Seed, Madison, WI) 6 d after planting. Seeds were germinated in potting soil (Fruhstorfer Erde Typ T containing 250–380 mg/L nitrogen, 220–330 mg/L P₂O₅, 330–500 mg/mL K₂O, pH 5.7–6.3; http://www.hawita-gruppe.de), and plants were kept in the greenhouse with a 16-/8-h-photo (15,000–20,000 lux)/-dark period at 28°/20°C and a humidity of 35%/75%. Ustilago maydis strains FB1 (a1b1), FB2 (a2b2; Banuett and Herskowitz, 1989), FB1Δmrb1, FB2Δmrb1 (Bortfeld et al., 2004), AB311 (a1b1) and AB312 (a2b2; Brachmann et al., 2003), and JF1 (a1mfa2bE1bW2; Farfsing et al., 2005) were grown at 28°C in yeast/peptone/Suc (Tsukuda et al., 1988), potato dextrose (PD; DIFCO Laboratories, Detroit), or complete medium (Holliday, 1974). Sporidial cultures after overnight growth in yeast/peptone/Suc medium were adjusted to a cell density of 5 × 10⁷/mL in distilled water. Plant inoculation was performed by injection of approximately 300 to 500 μL sporidial suspension through the stem and attached whorl approximately 2 cm above ground using a 26-gauge needle. Dikaryotic hyphae were harvested from solid charcoal (1% w/v) containing complete medium at room temperature 48 h after mixing compatible strains FB1 and FB2. Colletotrichum graminicola wild-type strain CGM2 (provided by Holger Deising; University of Halle-Wittenberg, Germany) was maintained at 22°C on oatmeal agar (50 g oatmeal, 12 g agar, 500 mL distilled water).

Infection of Detached Maize Leaves

Conidia of C. graminicola were collected from agar sections by flooding them in distilled water and were washed twice to remove germination inhibitors. Each 0.5 mL of either U. maydis sporidia (5 × 10⁶/mL) of strain JF1 or C. graminicola conidia (5 × 10⁶/mL) were spotted on the adaxial side of detached maize leaves (second leaf) cut from 6-d-old maize plants. Leaves were placed in a moisture chamber and incubated with a light/dark period of 16/8 h (10,000 lux) at 24°C/22°C. Infection was verified by staining with CBE (Brundrett et al., 1996). Untreated control leaves were incubated under the same conditions. The entire parts of each three identically treated leaves were used for subsequent extraction of total RNA.

DNA and RNA Procedures

DNA and RNA procedures were performed as described (Basse et al., 2000). Escherichia coli K12 strain DH5α (BRL, Gaithersburg, MD) and TOP10 (Invitrogen, Karlsruhe, Germany) were used as hosts for plasmid amplification. ³²P-labeled fragments for RNA-blot analysis were generated using the megaprime DNA labeling kit (Amersham-Pharmacia Biotech, Freiburg, Germany). For each RNA-blot analysis, total RNA was isolated from one experimental series and comparable amounts of RNA (5–10 μg) were loaded per lane. Detection and quantification of radioactive signals was done using a STORM PhosphorImager and the ImageQuant software (Molecular Dynamics, Sunnyvale, CA). The pCR2.1-TOPO, pCR4-TOPO, and pCR4Blunt-TOPO vectors (Invitrogen) were used for cloning of PCR products. The correctness of all PCR products was verified by sequencing. All other DNA manipulations followed standard procedures as described by Sambrook et al. (1989). Nucleotide/protein sequences were compared using BLASTX (Altschul et al., 1997) and PFAM (http://www.sanger.ac.uk/Software/Pfam). Prediction of signal sequences were made with program TargetP (www.expasy.ch).

Differential Display

Differential display and verification of differential fragments by Southern-blot hybridization was performed as described (Basse et al., 2000). Differential cDNA fragments were reamplified using the same primer combination and either cloned into the SmaI site of pUC19 (Amersham-Pharmacia Biotech) or the pCR2.1-TOPO vector. For subsequent Southern analysis cloned fragments were reisolated. Blots were prepared in duplicate and hybridized with the original ³⁵S-labeled PCR products from either control (noninfected) or tumor samples.

Isolation of cDNA Clones

Partial cDNA clones were isolated from Umi11 using a maize lambdaZAP-cDNA library kindly provided by Prof. Reinhard Kunze (Humboldt-University, Berlin) according to Stratagene's (La Jolla, CA) protocol, from Umi2 by 5′ RACE using the SMART kit (Roche, Mannheim, Germany), and from Umi5 and Umi6 by PCR using a cDNA library (Stratagene) as template representing cDNAs from infected maize tissue 2 and 5 dpi (strain mixture of FB1 and FB2; J. Farfsing, R. Kahmann and C.W. Basse, unpublished data). The complete 5′ cDNA sequences of Umi11 and Umi2 were determined by 5′ RACE (Invitrogen) using DNase-treated RNA from tumor tissue 6 dpi (strains FB1/FB2) and the nested gene-specific primers 5′-GCATGTCTCCCCCGTGTAGCACAGCGTG-3′/5′-CTTGGGCGTCGTGGCGAGGGAGCCATCC-3′ and 5′-GGGTATCATCTAGAATCGTGATGAAAGC-3′/5′-GTCTACAGTTCGTGCAAACTCTAGTACC-3′, respectively. To confirm the deduced ORF sequence of Umi2 the entire ORF was amplified from the same cDNA preparation using the primer combination (5′-ACCATGGCTGCCCCAACACTAACTGC-3′/5′-CATGAGTACCGGCTTCACATAAAGC-3′) and Platinum Pfu polymerase (CLONTECH, Heidelberg).

DNA Fragments for RNA-Blot Analysis

For RNA-blot analysis cloned fragments from differential display analysis were either isolated from pUC19 using EcoRI/PstI (Umi4, 5, 6, 12, Umr1) or from pCR2.1-TOPO using HindIII/XbaI (Umi7, 8, 9, 11, Umr2). Fragments corresponded to the regions indicated in Table I except for Umi5, Umi6, and Umi11, which corresponded to positions 230 to 636, 67 to 364, and 22 to 542 of the indicated cDNA regions. The Umi2 fragment was amplified from a tumor-specific cDNA preparation (see above) using the primer combination 5′-AGCACGAGGACACTGACATGG-3′/5′-CAGGAATGCTGCATTGTATAAGC-3′. Fragments of the maize actin gene MAc1 (accession no. J01238) and the maize PR-1 gene were amplified from genomic DNA using the primer combinations 5′-GGATTCAGGTGATGGTGTGAGC-3′/5′GCTAGGAACTAGGGACGTGATC-3′ and 5′-AGGCTAGCGTGCCTCCTAGCTCTGG-3′/5′-GGAGTCGCGCCACACCACCTGCGTG-3′, respectively. Cloned PCR fragments were reisolated from pCR-TOPO vectors using EcoRI.

RT-PCR Analysis

DNase (Roche)-treated RNA was reverse transcribed (Basse et al., 2000) with the primer TTGTACAAGCT₃₀VN and used as template for PCR with Taq polymerase (MBI Fermentas, St. Leon-Rot, Germany). Gene-specific primer combinations were used to determine expression of mig2-5 (m25gsa/m25gsb; Basse et al., 2002a) and Umi2 (5′-CCATATCTCACTACCAAAAGCATCCGATCC-3′/5′-GTTCCAACCCTTCGAAAAAGTGGTGCG-3′). The primers designed against the Umi2 sequence did not match the sequence of the homologous TPS5 gene. The ubiquitously expressed ip (Broomfield and Hargreaves, 1992) and maize histone H2B genes (Rasco-Gaunt et al., 2003; accession no. X57312) were amplified using the gene-specific primer combinations 5′-GTCGCTATTCAACGTCAGCAACGGTCTTCG-3′/5′-GAGCGAAAAGGTGTTCTCGAGCTTCTGTC-3′ and 5′-GAGAAGGCTCCGGCGGGGAAGAAGCCCAAG-3′/5′-AGGACGAGGCGCACCGAGGTCTGGATC-3′, respectively.

Isolation and Detection of DIMBOA

Tumors and control samples were collected from infected (FB1 × FB2) and mock-infected (FB1) leaf tissue, respectively, 6 dpi. Each sample was isolated from a different plant. Plant material was extracted as described (Frey et al., 1997) and subjected to HPLC (Beckman, Munich) on a C18 ultrasphere column (5 μm, 4.6 × 250 mm; Beckman) eluted with 1% (v/v) methanol, 10% (v/v) acetic acid in water for 0.6 min followed by a linear gradient to 20% (v/v) methanol in 20 min at a flow rate of 1 mL/min. Using this gradient DIMBOA eluted at 16.8 min.

DIMBOA/MBOA Inhibition Experiments

DIMBOA/MBOA (kindly provided by Prof. Dieter Sicker, University of Leipzig, Germany) and Trp were concentrated in water, and 50 μL of appropriate dilutions were dropped into wells of a culture plate (Greiner Bio-One, Kremsmünster, Austria) and mixed with 0.5 mL agarose-containing PD medium. U. maydis sporidial cultures were harvested from the logarithmic phase and each 5 μL of cultures resuspended in water, corresponding to 4 × 10⁴ cells, were subsequently spotted onto the solidified medium. Plates were incubated at 28°C for 24 h.

Microscopy

Microscopy and processing of images was performed as described (Basse et al., 2000). Samples were observed with differential interference contrast optics or under fluorescence microscopy (excitation/emission for GFP: 450–490/515–565 nm). All pictures were prepared using Adobe Photoshop (Adobe Systems, Mountain View, CA) and applying identical processing functions, with only those that could be applied equally to all pixels of the image being used.

Distribution of Materials

Upon request, all novel materials described in this publication will be made available in a timely manner for noncommercial research purposes.

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers AY647253, AY679127, AY679128, AY679129, and AY679130.