Transcriptomic analysis of Ustilago maydis infecting Arabidopsis reveals important aspects of the fungus pathogenic mechanisms

Abstract

Transcriptomic and biochemical analyses of the experimental pathosystem constituted by Ustilago maydis and Arabidopsis thaliana were performed. Haploid or diploid strains of U. maydis inoculated in A. thaliana plantlets grew on the surface and within the plant tissues in the form of mycelium, inducing chlorosis, anthocyanin formation, malformations, necrosis and adventitious roots development, but not teliospores. Symptoms were more severe in plants inoculated with the haploid strain which grew more vigorously than the diploid strain. RNA extracted at different times post-infection was used for hybridization of one-channel microarrays that were analyzed focusing on the fungal genes involved in the general pathogenic process, biogenesis of the fungal cell wall and the secretome. In total, 3,537 and 3,299 genes were differentially expressed in the haploid and diploid strains, respectively. Differentially expressed genes were related to different functional categories and many of them showed a similar regulation occurring in U. maydis infecting maize. Our data suggest that the haploid strain behaves as a necrotrophic pathogen, whereas the diploid behaves as a biotrophic pathogen. The results obtained are evidence of the usefulness of the U. maydis-A. thaliana pathosystem for the analysis of the pathogenic mechanisms of U. maydis.

Introduction

The use of alternative hosts to understand pathogenic processes is a strategy widely used since the earliest microbiological studies on human diseases. More recently, these studies extended to fungal infections, e.g., rabbits and mice have been used as hosts to study fungal human pathogens as Cryptococcus neoformans and Candida albicans^1,2 and even Drosophila melanogaster has served as alternative host for studying human mycoses.³ Additionally, model plants are used to study fungal pathogens of commercial interest. Thus Nicotiana benthamiana was employed as alternative host for mutants of Colletotrichum orbiculare, a pathogen of melon and cucumber⁴ and Brachypodium distachyon was used as alternative host of Magnaporthe grisea, a rice and barley pathogen.⁵ Additionally, in the pathosystems Cryptococcus-Arabidopsis and Cryptococcus-Eucalyptus, the human pathogen completed its sexual cycle.⁶

Similarly, our research group demonstrated that Ustilago maydis infected different plant species, unrelated to its natural host, Zea mays under axenic conditions,⁷ establishing the Ustilago maydis-Arabidopsis thaliana pathosystem as model to study some virulence aspects of the fungus.⁸ A. thaliana has the attractive characteristics of its small size, short life cycle and its complete genome sequence. Furthermore, Arabidopsis is a model for plant-pathogen interactions.^9,10

U. maydis is a pathogen of maize (Zea mays) and teozintle (Zea mays ssp parviglumis), where it completes its known sexual life cycle. This starts when two haploid yeast-like cells of compatible mating types fuse giving rise to a dikaryotic hypha that infects the plant. Inside the host, the mycelium undergoes morphological changes eventually giving rise to teliospores that accumulate within tumors induced by the pathogen. Teliospores germinate to form haploid basidiospores that reinitiate the life cycle.^11,12 Interestingly, we demonstrated that U. maydis performs a completely different sexual life cycle with the surprising formation of basidiocarps, when incubated under defined environmental conditions.¹³

Considering the complexity of the sexual-pathogenic process of U. maydis in maize, the Ustilago maydis-Arabidopsis pathosystem appears as an attractive alternative, since no sexual cycle occurs, and haploid strains are pathogenic. In this study we analyzed the infection of Arabidopsis by haploid or diploid strains of U. maydis and compared the transcriptomic differences of both strains to identify factors that explained the specificity of maize infection by dikaryons or diploids and to define similarities and differences in the infection of Arabidopsis and maize.

Results

Symptoms in plantlets infected with U. maydis strains

Haploid or diploid strains of U. maydis infected Arabidopsis plantlets as described,⁸ but the haploid induced more severe symptoms. Aerial mycelium of the haploid strain developed at the inoculation sites after 24 h, spreading to other regions (Fig. 1A and B, arrows), necrotic areas developed 4 d post-infection (dpi) and tissue detachment at 8 dpi (Fig. 1B and C, respectively, arrows). Mycelium developed extensively within plant tissues (Fig. 1I), possibly penetrating through stomata (Fig. 1H, red arrow). The haploid strain also induced severe alterations in the root system at 20 dpi (Fig. 1G). On the other hand, plantlets infected with Uid1 developed scant mycelium at early periods (Fig. 1D and E) and necrosis points only after 8 dpi (Fig. 1F, arrow).

Figure 1. Appearance of A. thaliana plantlets infected with U. maydis strains. General aspect of plants inoculated with the FB2 haploid strain (A–C), or the Uid1 diploid strain (D–F) at 2, 4 or 8 d post-infection respectively. Notice in (B) the necrotic area (arrow), in (C) necrotic tissue detachment (arrow) and in (F) necrotic tissue points in the infected plantlets. (G) Severe alteration in the root system in haploid-infected plantlets 20 dpi. (H) Mycelium of the haploid strain entering through stomata (red arrow).(I) mycelium of the haploid strain growing within the infected tissue. (J–L) Aspect of roots and leaves of control not-inoculated plantlets. Scale bar in (H) and (I), 20 μm.

Effect of U. maydis inoculation on growth of Arabidopsis plantlets

Growth of Arabidopsis plantlets infected with the haploid was reduced approximately from 50 to 90 per cent after 4 dpi, whereas diploid-infected plantlets were similar to controls until 12 dpi, before slowing down. Dry weight (DW) of all plantlets was similar until 4 dpi, but thereafter DW of haploid-inoculated plantlets remained constant, while that of diploid-inoculated plantlets remained similar to controls until 8 dpi, after which they showed similar growth as the control plantlets (Fig. 2).

Figure 2. Growth of A. thaliana plantlets infected with U. maydis strains. (A) Plantlets length and (B) plantlets dry weight. Dotted lines with circles, plantlets infected with the haploid strain. Solid lines with squares, plantlets infected with the diploid strain. Dotted lines with triangles, control plantlets that received sterile distilled water. Results from three independent experiments with six plants in each one. Results expressed in cm or g per plantlet respectively in each graph. Bars represent standard error values.

Growth of U. maydis within Arabidopsis plantlets

Growth of U. maydis was measured by ergosterol and chitin, compounds that are present in fungi, but absent in plants. Amounts of chitin and ergosterol in plants infected with either strain were similar at 4 dpi, but afterwards, the levels of both compounds were higher in haploid-infected plants, whereas they decreased in diploid-infected plants, suggesting curtailment of fungal growth by the plant, something not occurring with the haploid strain. As expected, in the un-inoculated plants, no ergosterol or chitin were detected (Fig. 3).

Figure 3. Development of U. maydis within infected A. thaliana plantlets. (A) Ergosterol determination. (B) Chitin determination. Empty circles with dotted lines, plantlets infected with the haploid. Empty squares with solid lines, plantlets infected with the diploid. Empty triangles with dotted lines, control plantlets that received sterile distilled water. Results from three independent experiments with 100 plant batches in each one. Results expressed as ng ergosterol or mg GlcNAC per plantlet. Bars represent standard error values.

Differential gene expression in U. maydis strains infecting Arabidopsis

Analysis of the genes differentially expressed in Ustilago strains during Arabidopsis infection was compared with their expression in yeast-like cells grown in liquid culture. As considered significant in most microarray analyses, a two-fold change up or down was used to consider differential expression of genes. The results obtained revealed that in the haploid 2,636 genes were differentially expressed at 1 dpi, 1,294 up-regulated and 1,342 down-regulated. In the diploid the corresponding numbers were 2,389; 1,170 and 1,219, respectively. In both cases the numbers remained almost constant during the following dpi (Table 1). These numbers reveal the extreme changes occurring when U. maydis changes its mode of life from saprophytic to pathogenic. The total number of genes differentially expressed along the infection process were 3,537; 1,703 up-regulated and 1,834 down-regulated in the haploid and 3,299; 1,621 and 1,678 respectively in the diploid; genes commonly regulated in both strains were respectively, 696, 407 and 289. Table S1 describes the genes with the highest difference in expression (at least a twenty-fold change up or down).

Table 1. Differential gene expression in U. maydis strains during infection of Arabidopsis thaliana plantlets.

Strain	Genes	Days post-infection
		1	2	4	8
Haploid (FB2)	Differential total	2,636	2,315	2,503	2,320
		Upregulated	1,294	1,132	1,216	1,152
		Downregulated	1,342	1,183	1,287	1,168
Diploid (Uid1)	Differential total	2,389	2,420	2,162	2,162
	Upregulated	1,170	1,203	1,084	1062
	Downregulated	1,219	1,217	1,078	1,100

Functional classification of differential genes

Functional grouping of the differentially expressed genes in either strain gave similar data (Fig. 4). The categories with higher gene numbers were related to metabolism, energy and metabolism regulation (19.1% in the haploid and 18.7% in the diploid of the total differential genes), transcription, protein synthesis and fate (respective values of 14.9% and 14.1%). The categories with lower gene numbers were interaction with the environment (respectively, 5.3% and 5.4%) and cell differentiation (respectively, 2.9% and 2.9 %) (Fig. 4).

Figure 4. Functional categorization of U. maydis genes regulated during the infective process of Arabidopsis. (A) Differentially regulated genes in the haploid. (B) Differentially regulated genes in the diploid. Numbers represent the percentage of genes in a given functional category and text corresponds to the names of each category.

Classes of differentially regulated genes

Considering the high number of differentially regulated genes during Arabidopsis infection, we investigated their nature according to classes related to pathogenicity, some of which are regulated during maize infection.

Genes regulated by the bE/bW heterodimer

It has been described that U. maydis infection in maize by heterokarions or diploids, is regulated by a heterodimer made of the bE and bW gene products coming from the two sexually-interacting strains; this heterodimer acts as a master transcription factor for a number of genes ^11,14,15 and it has been hypothesized that it is also involved in the U. maydis biotrophic phase.¹⁶ Accordingly, we analyzed their regulation during Arabidopsis infection considering that only the diploid contains the heterodimer. PRF1, the gene directly regulating transcription of the heterodimer was repressed in the haploid, whereas no differential expression was observed in the diploid. Of the genes directly regulated by the heterodimer: CLP1 (required for intracellular hyphal proliferation), FRB52 (of unknown function) and RBF1,¹⁶ only RBF1 was up-regulated in the diploid (Table 2). In turn, Rbf1 regulates BIZ1, HDP1, HDP2 and FOX1 encoding transcription factors and two additional proteins: KPP6 and PCL12. HDP1 was up-regulated only in the haploid, FOX1 was up-regulated in the diploid and down-regulated in the haploid, whereas no differential expression of BIZ1 and HDP2 was observed in any strain. PCL12 and KPP6 were up-regulated in both strains (Table 2).

Table 2. Differential expression of genes regulated by the bE/bW heterodimer during U. maydis infection of A. thaliana.

Identity		Gene	Description	Regulation
				Haploid strain FB2	Diploid strain Uid1
um03172		Rbf1	Related to Zinc finger protein	-	up
um02438		Clp1	Related to clp1, essential for A-regulated sexual development	-	-
um01262		Frb52	DNA polymerase X-putative	up	up
um02331		Kpp6	MAP kinase	up	up
um02549		Biz1	b-induced zinc finger protein	-	-
um12024	um05762	Hdp1	Potential homeodomain transcription factor	up	-
um04928		Hdp2	Uncharacterized protein	-	-
um10529	um10529.2	Pcl12	b-dependently cyclin	up	up
um01523		Fox1	Transcriptional regulator	down	up

Genes belonging to 12 pathogenicity clusters

Expression of genes belonging to the twelve pathogenic clusters in the diploid strain was similar to the one observed in maize.^11,15 Other genes, especially those belonging to cluster 2A whose deletion increases virulence, were down-regulated in the haploid and up-regulated in the diploid as occurs in maize (Table S2).

Effector genes

Effectors are proteins secreted by plant pathogens that manipulate the physiology of their hosts to their benefit.¹⁷ In biotrophic fungi, these effectors accumulate in the zone of interaction with the host inducing the suppression of plant immune responses.¹⁸ Fox1, controlling the expression of effector-encoding genes in U. maydis,¹⁹ specifically PEP1, the EFF1 family and PIT2,^20-22 was up-regulated in the diploid and down-regulated in the haploid. PIT2 was down-regulated in the haploid and was not differentially expressed in the diploid; EFF1-5 to EFF1-11 were up-regulated in the diploid, whereas in the haploid EFF1-6 and EFF1-8 were not differentially expressed and EFF1-7 was down-regulated. EFF1-9, EFF1-10 and EFF1-11 were up-regulated but at most at two different dpi. In contrast, EFF1-1 to EFF1-4 and PEP1 were not differentially expressed in either strain (Table 3).

Table 3. Differential expression of genes encoding U.maydis effector proteins during Arabidopsis infection.

Identity	Gene	Regulation in Haploid (FB2)				Regulation in Diploid (Uid1)
		1 dpi	2 dpi	4 dpi	8 dpi	1 dpi	2 dpi	4 dpi	8 dpi
um01375	Pit2	2.0 down	2.2 down	-	-	-	-	-	-
um02135	Eff1-5	-	-	3.1 up	2.7 up	5.2 up	5.4 up	6.5 up	5.8 up
um02136	Eff1-6	-	-	-	-	2.6 up	2.7 up	-	2.1 up
um02137	Eff1-7	2.2 down	2.2 down	-	-	-	-	2.6 up	2.5 up
um02138	Eff1-8	-	-	-	-	4.0 up	4.4 up	3.3 up	4.2 up
um02139	Eff1-9	-	-	2.2 up	2.2 up	-	-	3.3 up	2.1 up
um02140	Eff1-10	-	-	2.1 up	3.3 up	2.6 up	-	-	-
um02141	Eff1-11	-	2.1 up	-	2.0 up	4.8 up	4.4 up	2.4 up	2.5 up

Genes encoding hydrophobins and repellent proteins

Hydrophobins and repellent proteins are involved in U. maydis virulence in maize. HUM2 has essential functions in maize infection and REP1 is involved in aerial hyphae formation.^23,24 We found that during Arabidopsis infection REP1, HUM2 and HUM3 were up-regulated in the haploid strain, particularly REP1 (39.3 of fold change at 1 dpi). In contrast, only REP1 and HUM2 were up-regulated in the diploid (not shown).

Genes encoding transcription factors

We identified 111 genes encoding potential transcription factors differentially expressed during Arabidopsis infection. In the haploid, 45 were up-regulated and 66 down-regulated. In the diploid 47 were up-regulated and 63 down-regulated. Seventy-two genes were up- or down-regulated in common in both strains (27 up- and 45 down-regulated). In the haploid, gene um02301 was highly over-expressed (25.5-fold change) and um06283 was highly repressed (15.1-fold change). In the diploid, genes um00841 and um05966 were highly over-expressed (9.0 and 7.7 of fold change, respectively), whereas um06283 and um04208 were severely repressed (8.7 and 11.1 of fold change, respectively) (Table S3). The importance of regulation of transcription factors during maize infection is known,^16,25 in agreement with our data on Arabidopsis infection.

Other genes reported as involved in maize infection

Around fifty genes have been described to be involved in maize infection by U. maydis, as deduced from the phenotype of their corresponding mutants. In Arabidopsis infection we identified that 21 of them were differentially regulated. In the haploid, 12 were up-regulated and 9 were down-regulated and in the diploid 8 were up-regulated and 10 down-regulated (Table 4).

Table 4. Regulation of Ustilago maydis genes described as important for virulence in maize, during Arabidopsis infection.

Identity		Description	Reference	Fold change during Arabidopsis thaliana infection
				Days post-infection with the haploid strain FB2				Days post-infection with the diploid strain Uid1
				1	2	4	8	1	2	4	8
um00641	um11450	Hgl1; Hgl1p, required for dimorphism and teliospore formation	Deletion67	4.4 down	5.2 down	-	-	3.0 down	3.9 down	3.1 down	-
um01374		Pit1	Deletion21	-	2.2 up	3.9 up	3.1 up	-	-	2.2 up	2.2 up
um01597		Hap2	Deletion68	5.2 down	2.8 down	2.6 down	-	2.9 down	2.2 down	-	-
um01947		Related to cytochrome-c peroxidase precursor	Deletion69	2.6 down	16.5 down	14.8 down	48.2 down	-	-	3.0 down	3.6 down
um02374		Srt1; related to monosaccharide transporter	Deletion70	3.1 down	2.7 down	3.2 down	-	3.9 down	3.5 down	-	-
um02377		Probable cytochrome c peroxidase precursor	Deletion69	-	-	-	-	3.7 up	3.3 up	2.2 up	2.9 up
um03280	um10828	Tup1; probable TUP1-general transcription repressor	Deletion28	-	-	-	-	2.2 down	-	-	-
um03305		Ubc3; MAP kinase	Deletion71	3.5 up	3.1 up	3.6 up	3.4 up	2.3 up	2.3 up	-	-
um03315		Ukb1; serine/threonine protein kinase b-related Ukb1	Deletion72	3.1 up	2.4 up	2.8 up	2.3 up	-	-	-	-
um04258.2	um04258	Ubc4 - MAPKK kinase	Deletion73	-	2.0 down	-	-	-	-	-	-
um04474		Gpa3; guanine nucleotide-binding protein α-3 subunit	Deletion74	2.3 up	2.8 up	2.8 up	4.3 up	3.2 up	3.3 up	3.9 up	3.8 up
um04580		O-mannosylation	Deletion75	3.3 up	-	2.3 up	2.0 up	2.5 up	2.0 up	-	2.0 up
um05261		Ubc2; MAP kinase pathway-interacting protein	Deletion76	2.9 down	2.7 down	2.8 down	2.0 down	4.2 down	4.2 down	3.6 down	2.5 down
um05433		Pmt4; probable PMT4; dolichyl-phosphate-mannose–protein O-mannosyltransferase	Deletion44	3.3 up	2.8 up	3.1 up	2.6 up	-	-	-	-
um05818		Probable chimeric spermidine synthase/saccharopine reductase	Deletion77	-	-	2.4 up	2.7 up	2.0 down	-	-	2.0 down
um05850	um05850.2	Polyamine oxidase (PAO)	Deletion78	3.5 up	2.4 up	2.4 up	2.2 up	2.3 up	-	2.2 up	2.0 up
um10417	um04252	Nit2; related to transcription factor ScGATA-6	Deletion29	-	-	-	-	2.7 down	2.7 down	2.2 down	-
um10672	um10672.2	Peroxidases	Deletion69	2.7 down	-	2.3 down	2.3 down	4.5 down	5.2 down	4.1 down	5.7 down
um10792	um00999	Related to S-adenosylmethionine decarboxylase (spe-2)	Deletion79	3.0 down	2.0 down	2.1 down	-	-	-	-	-
um10803	um01516	Sql2; guanyl nucleotide exchange factor Sql2	Deletion80	2.7 down	2.6 down	2.5 down	2.2 down	2.1 down	2.5 down	-	-
um11410	um02513	Crk1; Cdk-related kinase 1	Deletion81	-	2.1 up	2.1 up	2.5 up	-	-	-	2.3 up
um11453	um00648	Ubc5; probable UBC5 - E2 ubiquitin-conjugating enzyme	Deletion73	-	-	-	-	2.5 up	3.1 up	-	-
um12033	um06025	Rop1 HMG-box transcription factor (C-terminal fragment)	Deletion82	-	-	3.1 up	-	-	-	-	-
um04456	um10537	Adr1 - protein kinase A, catalytic subunit	Described83	3.3 up	3.2 up	2.9 up	2.5 up	-	-	-	-
um12024	um05762	hdp1	Described16	-	-	5.6 up	2.0 up	-	-	-	-

Genes involved in the synthesis of structural polysaccharides

In the haploid, CHS1 one of the genes encoding chitin synthases, was up-regulated during the whole infection period, whereas CHS3 was down-regulated, but only at 8 dpi. In the diploid, no CHS gene was differentially expressed. Genes involved in the synthesis of β-1,3- or β-1,6-glucans behaved differently; GLS encoding β-1,3-glucan synthase was constitutively expressed in both strains, but of the genes involved in β-1,6-glucan synthesis, CWH41 and ROT2 homologs and three KRE6 homologs were up-regulated (at least at one dpi), whereas three other homologs of KRE6 were down-regulated in the haploid. In the diploid, two KRE6 homologs were up-regulated and two down-regulated (Table 5). Genes encoding chitin deacetylases (CDAs), enzymes involved in chitosan synthesis, were up-regulated in both strains, but their fold changes were greater in the haploid. In contrast, three other CDA genes were down-regulated in the haploid and only two of them in the diploid (Table 5).

Table 5. Differential expression of genes involved in the synthesis of structural polysaccharides of U.maydis during Arabidopsis infection.

Identity	Gene	Subcelular localization of protein	Haploid (FB2)				Diploid (Uid1)
			1 dpi	2 dpi	4 dpi	8 dpi	1 dpi	2 dpi	4 dpi	8 dpi
um10718	CHS1	Cell wall	3.2 up	3.4 up	2.1 up	2.3 up	-	-	-	-
um10120	CHS3		-	-	-	2.3 down	-	-	-	-
um01143	CDA		-	2.5 down	-	3.8 down	3.1 down	3.0 down	2.1	2.0 down
um01788			2.6 down	2.2 down	-	-	2.5 down	2.8 down	-	-
um02019			-	-	2.7 down	-	-	-	-	-
um05792			-	2.7 up	5.4 up	8.1 up	2.6 up	-	2.4 up	-
um11922			24.2 up	31.0 up	29.0 up	27.1 up	10.9 up	9.9 up	15.1 up	12.0 up
um01639	GLS		-	-	-	-	-	-	-	-
um11723	CWH41a	Endoplasmic Reticulum	3.4 up	3.1 up	2.7 up	2.7 up	-	-	-	-
um04405	ROT2a		2.1 up	2.1 up	2.0 up	-	-	-	-	-
um00857	KRE6/SKN1a	Golgi Apparatus	2.0 up	-	2.5 up	-	-	-	-	-
um03569			6.0 down	4.3 down	7.9 down	5.6 down	-	-	-	-
um05718			3.4 down	3.7 down	3.0 down	3.2 down	2.2 down	2.1 down	-	-
um05807			3.3 up	5.2 up	-	-	-	-	-	-
um05809			18.9 up	27.0 up	8.1 up	10.1 up	3.7 up	3.9 up	6.8 up	4.9 up
um05811			4.0 down	10.3 down	7.9 down	2..5 down	5.4 down	4.8 down	3.4 down	3.2 down

^a Homologs to Saccharomyces cerevisie genes.

Genes involved in polysaccharide biosynthesis precursors

One gene encoding a hexokinase and one encoding a N-glucose acetyl transferase involved in the synthesis of uridine diphosphate GlcNAc (UDP-GlcNAc) were up-regulated in both strains, whereas a gene encoding a phosphoacetylglucosamine mutase was down-regulated in both strains. Genes involved in the synthesis of uridine diphosphoglucose (UDPGlc) and guanosine diphospho manose (GDPMan) were also differentially expressed, thus, the gene encoding UDP-glucose-hexose-1-phosphate uridylyltransferase was up-regulated in the diploid strain and the gene encoding mannose-1-phosphate guanylyltransferase showed a decrease in the level of transcription in the same strain (not shown).

Genes involved in the synthesis of glycoproteins

In the haploid, almost all the genes involved in the synthesis of the N-linked glycan moiety of glycoproteins were up-regulated, with the exception of SEC59, encoding a dolichol kinase, that was down-regulated, same as happened in the diploid, where only a few genes were up-regulated (Table 6). In contrast, fewer genes involved in the synthesis of the O-glycan moiety of glycoproteins were differentially regulated. Of PMT genes encoding enzymes that transfer the first mannose to Ser/Thre residues of proteins, only PMT4 was up-regulated in the haploid and genes involved in further addition of mannosyl units (KRE2 homologs) were up-regulated in both strains (Table 6).

Table 6. Differential expression of U.maydis genes involved in N- and O glycosylation during Arabidopsis infection.

Sequence identity	Homolog genea	Function	Haploid (FB2)				Diploid (Uid1)
			1 dpi	2 dpi	4 dpi	8 dpi	1 dpi	2 dpi	4 dpi	8 dpi
um10484	ALG7	Oligosaccharide synthesis	5.4 up	4.1 up	2.7 up	2.5 up	8.1 up	4.6 up	4.0 up	4.3 up
um11771	ALG14		3.1 up	-	-	-	2.5 up	2.1 up	-	-
um11547	ALG1		2.3 up	-	2.2 up	-	-	-	-	-
um05209	ALG11		2.1 up	2.2 up	2.3 up	-	-	-	-	-
um02433	RFT1		2.5 up	2.6 up	2.1 up	-	-	-	-	-
um11341	ALG12		4.2 up	3.8 up	3.7 up	2.8 up	6.5 up	5.3 up	4.7 up	3.6 up
um04779	SEC59		2.7 down	2.4 down	2.2 down	2.1 down	2.4 down	2.5 down	2.1 down	-
um01231	ALG5		2.3 up	2.2 up	2.3 up	2.0 up	2.0 up	2.0 up	2.1 up	2.3 up
um05547	ALG10/DIE2		-	-	-	-	2.1 up	-	-	-
um01062	VRG4		2.1 up	-	-	-	-	-	-	-
um04198	OST3	Oligosaccharide transference	3.8 up	2.8 up	3.6 up	2.8 up	2.8 up	2.9 up	2.2 up	2.0 up
um05293	STT3		3.5 up	3.8 up	3.6 up	3.6 up	2.4 up	2.2 up	2.5 up	2.3 up
um11723	CWH41/GLS1	Modification of oligosaccharide	3.4 up	3.1 up	2.7 up	2.7 up	-	-	-	-
um04405	ROT2		2.1 up	2.1 up	2.0 up	-	-	-	-	-
um01957	MNS1, MNL1/HTM1		4.0 up	1.6 up	3.0 up	2.1 up	7.3 up	4.6 up	8.1 up	5.7 up
um02227			2.8 up	2.5 up	2.6 up	2.7 up	4.9 up	5.6 up	6.8 up	5.5 up
um10494			3.7 up	2.3 up	2.5 up	2.1 up	-	-	-	-
um05433	PTM4	Transfer of first mannose to Ser/Thre	3.1 up	2.8 up	3.0 up	2.6 up	-	-	-	-
um01154	KRE2	Modification with additional mannose	3.8up	3.6 up	3.2 up	3.2 up	2.9 up	2.8 up	3.4 up	3.2 up
um01821			2.0 up	-	-	-	2.1 up	2.4 up	2.2 up	2.5 up

^a Homologs to S. cerevisiae genes

Genes GPI12, GPI14 and GUP1 involved in the synthesis of the GPI anchor of GPI proteins were up-regulated in both strains. Additionally, GPI17 and GPI18 were up-regulated and PER1 down-regulated in the haploid, whereas in the diploid GPI1 and GPI8 were up-regulated and GPI3 was down-regulated at different dpi (Table 7).

Table 7. Expression of U.maydis genes involved in GPI-anchor synthesis during Arabidopsis infection.

Identity	Homolog gene	Function	Process	FB2				Uid1
				1 dpi	2 dpi	4 dpi	8 dpi	1 dpi	2 dpi	4 dpi	8 dpi
um00213	GPI1	Modified of PI with GlcNAc (GPI-GnT complex)	Synthesis of GPI anchor from PI	-	-	-	-	3.4 up	3.2 up	2.6 up	2.5 up
um00302	GPI12	Deacetylation		-	2.0 up	-	-	2.5 up	2.0 up	2.0 up	-
um11347	GPI14	Transfer of mannose		2.9 up	4.5 up	3.3 up	4.1 up	-	-	2.0 up	2.4 up
um05554	GPI18	Transfer of mannose		2.4 up	2.9 up	2.6 up	2.3 up	-	-	-	-
um05709	GPI8		Catalytic site	-	-	-	-	2.3 up	2.0 up	-	-
um01565	GPI17			2.3 up	2.5 up	2.9 up	2.2 up	-	-	-	-
um04859	PER1	Removal of acyl group of GPI-anchor	Lipid Remodeling	4.5 down	3.9 down	4.3 down	3.9 down	-	-	-	-
um03003	GUP1	Lipid remodeling with longer fatty acids		5.3 up	4.1 up	4.9 up	4.5 up	2.8 up	2.9 up	3.0 up	3.1 up

Genes encoding GPI proteins and proteins of the secretome

CDA genes and other genes encoding GPI-hydrolytic enzymes were up-regulated in both strains (not shown), and numerous genes encoding secreted proteins classified as unknown or with degradative, synthetic, redox or non-enzymatic functions were differentially expressed in either strain. A greater number of genes encoding enzymes that degrade polysaccharides, lipids and proteins were up-regulated, in the haploid, as compared to the diploid strain (Table 8).

Table 8. Numbers of U.maydis genes encoding secreted proteins regulated during Arabidopsis infection.

Function		Regulate	Haploid (FB2)				Diploid (Uid1)
			1 dpi	2 dpi	4 dpi	8 dpi	1 dpi	2 dpi	4 dpi	8 dpi
Unknown		Up	54	50	62	57	61	72	62	76
		Down	29	37	27	32	17	20	30	22
Degradation	Polysaccharidases	Up	17	16	19	15	10	9	11	9
		Down	0	0	0	0	1	1	1	1
	Peptidases	Up	9	7	8	6	4	4	4	5
		Down	0	0	2	2	2	0	0	2
	Nucleases	Up	6	5	5	5	5	6	5	5
		Down	0	0	0	0	0	0	0	0
	Lipases and esterases	Up	4	5	6	8	4	5	3	5
		Down	0	1	1	0	1	1	1	1
	Phytases	Up	3	2	3	2	2	3	3	1
		Down	0	0	0	0	0	0	0	0
Synthesis		Up	3	4	5	4	4	2	4	2
		Down	1	2	3	3	1	4	2	3
Redox		Up	12	11	12	9	7	5	7	8
		Down	3	3	3	3	8	8	8	7
Non-enzymatic		Up	4	4	2	2	4	4	4	4
		Down	3	3	5	5	5	5	5	6

Discussion

Ustilago maydis is a useful model for understanding phenomena of fungal pathogenesis,^11,12 and Arabidopsis thaliana is an excellent model for understanding plant responses to pathogens.¹⁰ For these reasons, our working group established the experimental pathosystem Ustilago maydis-Arabidopsis thaliana, as model system for understanding some aspects of Ustilago pathogenesis.⁸ In this work we analyzed the transcriptome of U. maydis infecting Arabidopsis plantlets comparing a diploid and a haploid strain of the fungus, considering that diploids (as Uid1, used in this work) and heterokaryons, but not haploids, are infective in maize, but both infect Arabidopsis.

We observed a drastic alteration in gene expression in U. maydis during Arabidopsis infection, in numbers similar to maize infection,¹⁵ corresponding to more than one third of the whole genome. These numbers indicate that U. maydis drastically alters the levels of transcription of a great number of genes to cope with the changes taking place during its adaptation from a more or less comfortable environment where its nutritional necessities are satisfied, to the harsh conditions existing in the host, where it has to deal with the plant defenses and struggle for nutrients.

The damage caused to Arabidopsis plantlets by the haploid was noticeably more severe than that caused by the diploid and drastically inhibited plant growth, besides causing severe alterations in their roots. Growth of the haploid in the plant, measured by two specific parameters: ergosterol and chitin accumulation was also more abundant than growth of the diploid. These results reveal the higher virulence of the haploid, suggesting that it behaves as a necrotrophic parasite in Arabidopsis. These phytopathogens overturn the host defense mechanisms, destroy the plant,²⁶ and are resistant to host hypersensitive reactions.²⁷ In contrast, the diploid behaves as a biotrophic pathogen, as occurs in maize. This suggestion is supported by several pieces of evidence, e.g., the up-regulation in the diploid, contrasting with the haploid of RBDF1 (a master regulator required for all b-dependent processes) controlled by the bE/bW heterodimer and the genes that it regulates.¹⁶ Moreover, of the genes regulated by Rbf1, HDP1, involved in filamentous growth, was up-regulated only in the haploid, and FOX1 that regulates genes encoding effector proteins, was down-regulated in the haploid and up-regulated in the diploid. These data reveal, firstly, that genes encoding both transcription factors, as well as genes encoding Pcl12 and Kpp6, can be regulated by mechanisms alternative to Rbf1, and more important, that the opposite effect on HDP1 and FOX1 may be related to the different pathogenic behavior of both strains, biotrophic or necrotrophic. Additionally, a larger number of effectors, important for the biotrophic stage, were up-regulated in the diploid and were even down-regulated in the haploid. Further evidence to explain the different behavior of both strains and data of gene expression revealing homologies and differences in the pathogenic behavior of U. maydis in maize and Arabidopsis are described below.

In this sense, we must indicate that genes encoding transcription factors TUP1 and NIT2 involved in maize infection,^28,29 were also regulated during Arabidopsis infection, TUP1 being down-regulated in both strains and NIT2 repressed only in the diploid strain. Tup1 is repressed during maize infection and NIT2 disruption reduced virulence in maize. Behavior of homologs of genes encoding transcription factors related to virulence in different fungi was also significant: two homolog genes of HSF1 of C. albicans (that is up-regulated during formation of invasive mycelium)^30,31 and are up-regulated in U. maydis infecting maize, were differentially expressed only in the haploid, one up- and the other one down-regulated. Additionally, a homolog of SKN7 encoding a transcription factor required for virulence in C. albicans and C. neoformans^32,33 was up-regulated only in the haploid strain. These data provide evidence of the similarity in the pathogenic mechanisms of U maydis in maize and Arabidopsis.

Functional grouping identified only a limited number of genes involved in pathogenesis, as occurs in maize.³⁴ This result agrees with the knowledge that U. maydis possesses fewer pathogenesis genes than other phytopathogenic fungi,¹¹ possibly in relation to its biotrophic lifestyle. Among the factors involved in pathogenesis, genes encoding G proteins and transcription factors can be cited. The number of genes involved in cell rescue and defense that may be involved in response to stress and in detoxification was small, and interestingly, were mainly up-regulated in the haploid. Worth mentioning is that genes involved in response to stress, such as oxidases were repressed in the diploid, contrasting with the haploid where significant numbers of genes responding to oxidative stress were up-regulated, possibly as a defense mechanism. Other up-regulated genes mainly in the haploid strain are involved in homeostasis, cell migration, chemotaxis, mechanical stimulus perception and perception and response to nutrients. These functions are important for the formation of intracellular hyphae involved in acquisition of nutrients, signaling, communication and avoidance.³⁵

The process of Arabidopsis invasion, as occurs in maize, involves cell differentiation and dimorphic transition of yeast cells to invading hyphae. This process involves changes in expression of genes involved in cell wall biogenesis.³⁶ Accordingly, we analyzed the regulation of genes involved in this process. Some genes encoding glucanases or chitinases, probably involved in structural changes of the wall were differentially expressed during Arabidopsis infection. The higher number of CDA genes up-regulated in comparison to CHS genes, agrees with the observation that the hyphal surface of invasive biotrophic rust fungi, contains chitosan instead of chitin,³⁷ probably because chitosan, in contrast to chitin, lacks elicitor activity.

Expression of GLS encoding the single β-1,3-glucan synthase of Ustilago was constitutive during infection by both strains, agreeing with data from in vitro dimorphism or maize infection.³⁸ In contrast homologs of ROT2 and CWH41 from the pathway of β-1,6-glucans synthesis and N-glycosylation were up-regulated during Arabidopsis infection only in the haploid. Considering that mutants of the six homologs of KRE6/SKN1 in C. neoformans were avirulent to mouse,³⁹ the data reveal the importance of β-1,6-glucans synthesis in fungal pathogenesis, agreeing with the more aggressive behavior of the haploid strain.

Increased protein glycosylation is required for fungal-host interaction and virulence.⁴⁰ Accordingly, genes involved in the synthesis of N-glycans such as CWH41 and ROT2 and MSN1 and VRG4, have been related to virulence.⁴¹ In this regard, the observation that three MSN1 homologs from U. maydis were up-regulated in the haploid and only two in the diploid and that the homolog of VRG4 was up-regulated in the haploid is relevant. Regarding O-glycans, it was reported that PMT4, a gene involved in their synthesis, is involved in C. albicans and C. neoformans virulence,^42,43 and that its U. maydis mutants were avirulent to maize.⁴⁴ Agreeing with these data, we observed that this gene was up-regulated only in the haploid, correlating again with its higher virulence.

GPI proteins are important components of fungal membranes and walls and among them Yps aspartyl proteases are important in Candida pathogenesis,⁴⁵ agreeing with the observation that a gene encoding a Yps was up-regulated in both Ustilago strains. Taking into consideration that genes involved in the synthesis of the glycosylphosphatidyl inositol moiety of GPI-proteins such as GPI12, GPI17, GPI8 are important in virulence of fungi and protozoa,^46-48 it was significant to observe that in U. maydis the homolog of GPI12 was up-regulated in both strains and that GPI17 was up-regulated only in the haploid and GPI8 only in the diploid.

Hydrophobins and repellent proteins are important in Ustilago infection of maize.^23,24,49 We observed that REP1 and HUM2 genes were up-regulated in both strains, REP1 to higher levels in the haploid, whereas HUM3 was up-regulated only in this strain, suggesting their role in Arabidopsis infection.

Finally, it must be indicated that a higher number of genes encoding hydrolytic enzymes, some degrading the plant cell wall, were up-regulated in the haploid. This result may help to explain the greater damage produced by the haploid in Arabidopsis plants, agreeing with the observation that biotrophic fungi secrete a reduced number of degradative enzymes⁵⁰ and supporting our hypothesis of the different behavior of haploid and diploid strains of U. maydis during Arabidopsis infection.

In conclusion, our data that provide evidence of the importance of a number of genes in Ustilago virulence and that the genetic machinery used in Arabidopsis infection is similar to that used in maize infection. According to the evidence presented, we propose the hypothesis that the haploid strain of U. maydis behaves in Arabidopsis as a necrotrophic pathogen, in contrast to the diploid that, as occurs in maize, behaves as a biotrophic agent. We attribute this different behavior to alterations in the expression of genes encoding virulence factors, degradation enzymes, effector genes and transcription factors under the control of the bE/bW heterodimer, that obviously is absent in the haploid strain.

Materials and Methods

Fungal and plant strains, culture media and growth conditions

Fungal strains. U. maydis wild type strain FB2 (a₂b₂)⁵¹ and the diploid strain Uid1 (a₁b₁Δpan/a₂b₂Δodc::Hyg^R)⁵² were maintained and grown as described by Ruiz-Herrera et al.³⁶Arabidopsis thaliana L. Landsberg erecta (Ler) plantlets were grown on MS synthetic medium according to Mendez-Morán et al.⁸ (see below).

Arabidopsis growth conditions

A.thaliana Landsberg erecta seeds were sterilized with chlorine gas. Open Eppendorf tubes containing the seeds were placed in a desiccator that contained a beaker with 100 mL of concentrated hydrochloric acid and 5 mL sodium hypochlorite. The desiccator was kept closed during 4–6 h. After of sterilization, the seeds were maintained at 4°C during 2 d. For plant growth, 80–100 seeds were placed over plates of sterile solid MS medium and incubated in a chamber at 25°C with photoperiods of 12 h.

Inoculation and measurement of plantlets growth

U. maydis strains were grown in shaken liquid MC at 28°C for 18 h. The cells were recovered by centrifugation at 1,000 g for 10 m and washed twice with sterile distilled water (SDW) by centrifugation. Finally the cells were suspended in 5 mL of SDW and cell concentration was determined with a Neubauer chamber. A. thaliana plantlets were inoculated with 1–2 μl of a cell suspension containing 10⁶ cells/mL on the vegetative apex, 6 d post-planting. Control plantlets received SDW only. Plantlets were incubated under controlled conditions in a growth chamber as described above. At intervals, some plants were observed by light microscopy, directly or stained with cotton blue-lactophenol and photographed. At different periods post-inoculation plantlets were recovered and their dry weight and stem length were measured.

Ustilago maydis growth in A. thaliana seedlings

The biomass of U. maydis in inoculated plantlets was determined by ergosterol⁵³ and chitin⁵⁴ measurements. N-acetylglucosamine (GlcNAC) was determined as described by Reissig et al.⁵⁵

Isolation of RNA and microarrays hybridization of microarrays

Total RNA from Ustilago cells (three experiments in triplicates) or Arabidopsis plantlets (four experiments with 150 plants each) was isolated using Trizol (Invitrogen) and purified with QIAGEN columns (Cat. 28104). RNA concentration was measured by its absorbance at 260 nm and its integrity observed by electrophoresis in denaturing agarose gels. cDNA synthesis and labeling and hybridization of the microarrays were performed by Roche NimbleGen Inc.

Design, image analysis, normalization and analysis of microarrays

The microarrays used were high density (one channel) arrays from NimbleGene, according to a design from Scott Gold (University of Georgia). Five different oligonucleotide probes per gene (60 nt in length) per duplicate represented each one of the 6,883 genes of the U. maydis genome. GenePix 4000B scan and associated software were used to scan the arrays. Roche NimbleScan software was used to import the scanned images and data extraction. Normalization was made with NimbleScan software using quantile⁵⁶ and the algorithm Robust Multi-array Analysis (RMA).⁵⁷ Microarray analyses were made with DNAStar ArrayStar software, comparing data from Arabidopsis infected plants against U. maydis grown in culture medium. Data obtained from plants that received SDW were used as controls. P-values obtained were adjusted by the false discovery rate (FDR) method.⁵⁸ p-values inferior to 0.05 were considered significant. A two-fold change up or down was used to consider differential expression of genes. The Functional Catalogue (FunCat)⁵⁹ was used for functional annotation of differentially expressed genes.

Bioinformatic searches

Besides a classification of all regulated genes, in silico searches of specific genes previously reported in the following groups, were made: belonging to pathogenesis clusters,¹¹ encoding proteins from the secretome,^60,61 encoding hydrophobins or repellent proteins²⁴, regulated by the bE/bW heterodimer,^16,25 involved in the synthesis and organization of the cell wall,⁶² and those previously reported as important during maize infection (specific references cited in “Results”). Additionally, we performed in silico searches of genes encoding transcription factors using online programs UniProt,⁶³ KEGG,⁶⁴ Pfam,⁶⁵ SUPERFAMILY⁶⁶ and MIPS Ustilago maydis DataBase (http://mips.helmholtz-muenchen.de/genre/proj/ustilago/).

Glossary

Abbreviations:

dpi

days post-infection

DW

dry weight

FDR

false discovery rate

FunCat

functional catalogue

RMA

robust multi-array analysis

SDW

sterile distilled water

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.