The filamentous fungus Alternaria alternata includes seven pathogenic variants (pathotypes), which produce different host-selective toxins and cause disease on different plants. The Japanese pear, strawberry and tangerine pathotypes produce AK-toxin, AF-toxin and ACT-toxin, respectively, which have a common structural moiety, 9,10-epoxy-8-hydroxy-9-methyl-decatrienoic acid (EDA).
Here, we identified a new gene, AKT7 (AK-toxin biosynthetic gene 7), from the Japanese pear pathotype, which encodes a cytochrome P450 monooxygenase and functions to limit AK-toxin production.
AKT7 homologs were found in the strawberry pathotype, but not the tangerine pathotype. However, the strawberry pathotype homolog appeared to include a premature stop codon. Although the Japanese pear pathotype strain has multiple copies of AKT7, a single-copy disruption resulted in mutants with increased production of AK-toxin and EDA. AKT7 overexpression in the three pathotypes caused marked reductions of toxin and EDA production, suggesting that Akt7 catalyzes a side reaction of EDA or its precursor. AKT7 overexpression caused reduced virulence in these pathotypes. We also found that AKT7 transcripts predominantly include misspliced mRNAs, which have premature stop codons.
Our observations suggest that the AK-toxin production required for full virulence is regulated in a complex way by the copy number and intron information content of AKT7.
Filamentous fungi produce various secondary metabolites that are not necessary for normal growth or development, but may contribute to the fitness of the organisms in their natural environment (Keller et al., 2005; Howlett, 2006; Hoffmeister & Keller, 2007; Brakhage, 2013). Fungal secondary metabolites include substances involved in disease interactions with plants or animals. A number of plant pathogenic fungi produce secondary metabolite toxins that can damage plant tissues (Howlett, 2006; Hoffmeister & Keller, 2007; Stergiopoulos et al., 2013). Toxins are often classified as host selective (host specific) or nonspecific (Howlett, 2006; Tsuge et al., 2013). Host-selective toxins are toxic only to host plants of the fungus that produces the toxin. By contrast, nonspecific toxins can affect many plants regardless of whether they are a host or nonhost of the pathogen producing them. Host-selective toxins produced by fungal plant pathogens can act as effectors controlling pathogenicity or virulence in certain plant–pathogen interactions (Markham and Hille, 2001; Wolpert et al., 2002; Howlett, 2006; Tsuge et al., 2013).
There are now seven known diseases caused by the filamentous fungus Alternaria alternata in which host-selective toxins are responsible for pathogenicity; loss of toxin production results in loss of pathogenicity of these pathogens (Thomma, 2003; Tsuge et al., 2013). Alternaria alternata is one of the most cosmopolitan fungal species and is generally saprophytic (Rotem, 1994). Thus, the pathogens of this species have been defined as pathotypes of A. alternata (Tsuge et al., 2013).
Host-selective toxins from A. alternata pathotypes are diverse in structure (Tsuge et al., 2013). However, AK-toxin of the Japanese pear pathotype, AF-toxin of the strawberry pathotype and ACT-toxin of the tangerine pathotype have a common structural moiety, 9,10-epoxy-8-hydroxy-9-methyl-decatrienoic acid (EDA) (Fig. 1) (Nakashima et al., 1985; Nakatsuka et al., 1986; Kohmoto et al., 1993). The Japanese pear pathotype causes black spot on a narrow range of susceptible Japanese pear (Pyrus pyrifolia) cultivars; the strawberry pathotype causes Alternaria black spot on Japanese strawberry (Fragaria × ananassa) cv Morioka 16; the tangerine pathotype causes brown spot of tangerines (Citrus reticulata) and mandarins (C. reticulata) (Tsuge et al., 2013). Interestingly, in laboratory tests, the strawberry and tangerine pathotypes also show pathogenicity on Japanese pear cultivars susceptible to the Japanese pear pathotype (Maekawa et al., 1984; Kohmoto et al., 1993). This host range corresponds to the toxicity of AF-toxins and ACT-toxins. Most strains of the strawberry and tangerine pathotypes produce two related toxins, AF-toxins I and II in the case of the strawberry pathotype and ACT-toxins I and II in the case of the tangerine pathotype (Maekawa et al., 1984; Nakatsuka et al., 1986; Kohmoto et al., 1993) (Fig. 1). AF-toxin I is toxic to both strawberry and pear; toxin II is toxic only to pear (Maekawa et al., 1984). ACT-toxin I is toxic to both citrus and pear; toxin II is highly toxic to pear and slightly toxic to citrus (Kohmoto et al., 1993).
We previously isolated cosmid clones encoding part of the AKT (AK-toxin biosynthetic gene) cluster from the Japanese pear pathotype (Tanaka et al., 1999; Tanaka & Tsuge, 2000). By Southern blot analysis, AKT homologs were also detected in the strawberry and tangerine pathotypes, but not in other pathotypes or nonpathogenic strains of A. alternata (Tanaka et al., 1999; Masunaka et al., 2000; Tanaka & Tsuge, 2000; Hatta et al., 2002). Structural analysis of the cosmid clones encoding the AKT homologs in the strawberry and tangerine pathotypes, named AFT (AF-toxin biosynthetic gene) and ACTT (ACT-toxin biosynthetic gene), respectively, showed that the respective gene pairs of these three pathotypes share > 85% nucleotide identity (Hatta et al., 2002; Ruswandi et al., 2005; Miyamoto et al., 2008, 2009). This result reveals that these three pathotypes have conserved the genes required for EDA biosynthesis. We also identified genes unique to each pathotype within the AKT, AFT and ACTT clusters, which indicated that the gene clusters consist of pathotype-specific genes as well as genes common to the three pathotypes (Ito et al., 2004; Ajiro et al., 2010; Miyamoto et al., 2010). Structural and functional analysis of the toxin biosynthetic genes showed that the strains of the three pathotypes have multiple copies of functional or nonfunctional homologs of each gene in their genomes (Tanaka et al., 1999; Tanaka & Tsuge, 2000; Hatta et al., 2002; Ruswandi et al., 2005; Miyamoto et al., 2008, 2009, 2010; Ajiro et al., 2010). Thus, the genomic regions controlling biosynthesis of the toxin are not simple genetic loci, but instead are large, complex regions of DNA resulting from extraordinary duplication and recombination events.
Here we report the characterization of a new gene, AKT7 (AK-toxin biosynthetic gene 7), identified in the AKT cluster of Japanese pear pathotype strain 15A. AKT7 may encode a cytochrome P450 monooxygenase, and 15A has multiple copies of AKT7. Strawberry pathotype strains also possessed AKT7 homologs, but the homologs appeared to be pseudogenes with a premature stop codon. No AKT7 homolog was detected in tangerine pathotype strains. In this study, we analyzed the function of AKT7 in AK-toxin biosynthesis.
Materials and Methods
Fungal strains and genomic libraries
Strains of A. alternata used in this study are listed in Supporting Information Table S1. Strain 15A (ATCC90610) of the Japanese pear pathotype was the source of AKT7. The others were used for analysis of the distribution of AKT7 homologs, and strain NAF8 (MAFF655016) of the strawberry pathotype and strain SH20 of the tangerine pathotype were used for AKT7-1 overexpression experiments. Strains were routinely maintained on potato dextrose agar (PDA; Difco, Detroit, MI, USA). Genomic cosmid libraries of 15A, NAF8 and SH20 have been previously described (Kimura & Tsuge, 1993; Hatta et al., 2002; Miyamoto et al., 2008).
Nucleic acid manipulations
Isolation of total DNA and RNA from A. alternata and Southern blotting were performed as previously described (Tanaka et al., 1999; Harimoto et al., 2007). DNA was transferred to Hybond N+ nylon membranes (GE Healthcare Life Sciences, Piscataway, NJ, USA), then hybridized to AKT7-1 probes and detected using the Gene Images Random-Prime Labelling and Detection Reagents (GE Healthcare Life Sciences).
PCR was carried out using Ex Taq polymerase (Takara Bio, Shiga, Japan). Reverse transcription–polymerase chain reaction (RT-PCR) was performed using RNA PCR Kit ver. 2.1 (Takara Bio). The PCR and RT-PCR products were cloned into plasmid pGEM-T Easy (Promega, Madison, WI, USA) to determine their sequences.
DNA sequences were determined using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Warrington, UK) and an automated ABI PRISM 3100 genetic analyzer (Applied Biosystems) according to the manufacturer's instructions. DNA sequences were analyzed with the Blast search algorithm (Altschul et al., 1997). Alignment of nucleotide and amino acid sequences was made with the Clustal W program (Thompson et al., 1994).
Real-time quantitative RT-PCR was performed in a LightCycler Quick System 350S (Roche Diagnostics, Indianapolis, IP, USA) using the One Step SYBR RT-PCR Kit (Takara Bio) as previously described (Harimoto et al., 2007). The PCR primer pairs 7re-f1/7re-r1 and 7re-f2/7re-r2 (Table S2) were designed to amplify an AKT7-1 cDNA fragment of c. 150 bp using the primer design software in DNAsis Pro Ver. 2.2 (Hitachi Software Engineering, Yokohama, Japan). Primers were also designed for the actin gene (Actin-f/Actin-r) as a constitutively expressed endogenous control (Table S2). The expression level of AKT7-1 was normalized to that of the actin gene (value = 1). All calculations and statistical analyses were carried out as previously described (Harimoto et al., 2007).
The integrative transformation vector pSH75, which carries the hygromycin B phosphotransferase gene (hph) fused to the Aspergillus nidulans trpC promoter and terminator (Mullaney et al., 1985; Kimura & Tsuge 1993), was used for transformation of A. alternata.
The AKT7-1 disruption vector pGDAKT7 was constructed by cloning the 5′ and 3′ regions of AKT7-1 into pSH75 (Fig. S1a). The 5′ and 3′ regions of AKT7-1 were amplified from cosmid pcAKT-3 DNA by PCR using the primer pairs 7f1/7r1 and 7f2/7r2 (Table S2). Primers 7f1, 7r1, 7f2 and 7r2, respectively, have BglII, EcoRV, XbaI and XhoI sites. The amplified DNAs from the 5′ and 3′ regions were, respectively, digested with BglII-EcoRV and XbaI-XhoI, then cloned into the corresponding BglII-EcoRV and XbaI-XhoI sites of pSH75 to make pGDAKT7 (Fig. S1a).
The AKT7-1 overexpression vector pOEAKT7 was made by cloning AKT7-1 Type 1 cDNA between the A. nidulans trpC promoter and terminator of plasmid pEC2 (Inoue et al., 2002). AKT7-1 cDNA was amplified from total RNA of 15A with the primer pair 7oe-f/7oe-r (Fig. S2a, Table S2). The primer 7oe-f contains a BamHI site fused to the initiation codon; the primer 7oe-r contains an EcoRI site fused to the stop codon of AKT7-1. RT-PCR products were digested with BamHI and EcoRI and cloned into the BamHI-EcoRI site of pEC2, and pOEAKT7, carrying the trpC promoter-AKT7-1 Type 1 cDNA-trpC terminator construct, was selected (Fig. S2a).
All the PCR products that were cloned in these vectors were sequenced to confirm that no nucleotide substitution had occurred during amplification.
Protoplast preparation and transformation of A. alternata were performed by previously described methods (Tanaka et al., 1999). Transformants carrying the hph cassette were selected on regeneration media containing 100 μg ml−1 hygromycin B (Wako Pure Chemicals, Osaka, Japan).
Test for toxin production, pathogenicity and vegetative growth
To test for toxin production, strains of the Japanese pear and strawberry pathotypes were grown statically in 10 ml of potato dextrose broth (PDB; Difco) in 30-ml Erlenmeyer flasks at 25°C for 7 d; strains of the tangerine pathotype were grown statically in 10 ml of modified Richards' medium (1% KNO3, 0.5% KH2PO4, 0.25% MgSO4, 0.002% FeCl3, 0.0005% ZnSO4 and 2.5% glucose) in 30-ml Erlenmeyer flasks at 25°C for 20 d. To assay the toxicity of culture filtrates, Japanese pear (Pyrus pyrifolia var. culta Rehd.) cv Nijisseiki and strawberry (Fragaria × ananassa Duchesne ex Rozier) cv Morioka 16 were used. Leaves were wounded slightly, treated with culture filtrates and incubated in a moist box at 25°C for 24 h. Toxins and EDA in culture filtrates of the three pathotypes were quantified by reverse-phase high-performance liquid chromatography (HPLC) (Hayashi et al., 1990; Ito et al., 2004). Culture filtrate was injected into a Develosil ODS-7 column (Nomura Chemical, Seto, Japan) equipped with an LC-10AT pump (Shimadzu Corporation, Kyoto, Japan) and analyzed using an SPD-10A UV-VIS detector (Shimadzu Corporation). Elution of AK-toxin, AF-toxin, ACT-toxin and EDA, respectively, was with 45%, 38%, 38% and 20% acetonitrile acidified with 1% acetic acid at a flow rate of 1.0 ml min−1. Toxins were detected by monitoring absorbance at 290 nm for AK-toxin, AF-toxin and ACT-toxin and at 298 nm for EDA.
Oatmeal sucrose agar medium was used for preparation of conidia as previously described (Harimoto et al., 2008). A conidial suspension (c. 5 × 105 conidia ml−1) was spray-inoculated on leaves, and the leaves were incubated in a moist box at 25°C for 24 h.
To test for vegetative growth of A. alternata, strains were grown on PDA at 25°C for 4 d. PDA blocks (3 mm in diameter) carrying mycelia were inoculated on PDA, minimum agar medium (MA) (Sanderson & Srb, 1965) and complete agar medium (CA) (Sanderson & Srb, 1965). After incubation at 25°C for 5 d, colony growth and morphology were recorded.
Structure of pcAKT-3
We previously analyzed the structure of two cosmid clones, pcAKT-1 and pcAKT-2, of strain 15A of the Japanese pear pathotype, which contain parts of the AKT gene cluster (Fig. S3) (Tanaka et al., 1999; Tanaka & Tsuge, 2000). In this study, we determined the nucleotide sequence of pcAKT-3 (32 351 bp; accession no. AB872924), which contains a different part of the AKT region, and identified five genes (AKT2-2, AKTR-3, AKT3-3, AKT6-1 and AKT7-1), a pseudogene (AFT9h) and two transposon-like sequences within the region (Fig. 2). We found a 12.8-kb region that was highly conserved between pcAKT-1 and pcAKT-3: 99.8% nucleotide sequence identity (Fig. 2). This region of pcAKT-3 encodes AKT2-2, AKTR-3 and AKT3-3, corresponding to AKT2-1, AKTR-1 and AKT3-1, respectively, in pcAKT-1.
AFT9h in pcAKT-3 is a homolog of AFT9-1, which was found in the cosmid clone pcAFT-2 of the strawberry pathotype strain NAF8 and may encode a polyketide synthase (Figs 2, S3) (Ruswandi et al., 2005). Although AFT9h reveals strong similarity (97.0% nucleotide sequence identity) to the corresponding region of AFT9-1, it has an incomplete open reading frame (ORF) as a result of a 1159-bp deletion at the 5′ region and two stop codons, indicating that AKT9h is a pseudogene (Fig. 2).
AKT6-1 in pcAKT3 encodes a 298-amino acid protein with strong similarity to ACTT6 identified from the ACTT cluster of the tangerine pathotype strain SH20 (Miyamoto et al., 2009): 89.0% nucleotide sequence identity and 87.2% amino acid sequence identity (Figs 2, S3). ACTT6 encodes a protein of the enoyl-CoA hydratase/isomerase family. SH20 has two copies of ACCT6, and disruption of both copies in SH20 results in loss of ACT-toxin production and hence pathogenicity (Miyamoto et al., 2009). In Southern blot analysis, ACTT6 homologs were detected from strains of the tangerine, Japanese pear and strawberry pathotypes, but not from other pathotypes (Miyamoto et al., 2009). We isolated a cosmid clone, pcAFT-3, of the strawberry pathotype strain NAF8, which contains an AKT6 homolog named AFT6-1 (Figs 2, S3). Sequence analysis of pcAFT-3 (accession no. AB872925) detected a 5.2-kb region highly conserved between pcAKT-3 and pcAFT-3 and a complete ORF of AFT6-1 encoding a 298-amino acid protein within this region (Fig. 2). This result indicates that ACTT6, AKT6 and AFT6 of the three pathotypes are orthologs involved in EDA biosynthesis.
Structure and expression of AKT7
AKT7-1 in pcAKT3 is a new gene, which has not been previously characterized. A Blast database search revealed that AKT7-1 may encode a cytochrome P450 monooxygenase (Fig. S4). Based on the alignment of putative proteins encoded by AKT7-1 with homologous proteins from other fungi, AFT7-1 was deduced to consist of six exons (72, 151, 60, 264, 366 and 653 bp) divided by five introns (56, 57, 62, 57 and 67 bp) and to encode a 522-amino acid protein (Figs 2, 3). The structures of enzymes of the cytochrome P450 superfamily indicate special conservation of the cysteine heme-iron ligand signature at the C-terminus of the protein (Nelson et al., 1996; van den Brink et al., 1998). Akt7-1 possesses the signature sequence of 10 amino acids, with the conserved cysteine residue at amino acid positions 445–454 (Fig. S4). A Blast database search at the website for the Fungal Cytochrome P450 Database (http://p450.riceblast.snu.ac.kr/) (Park et al., 2008) revealed that Akt7-1 has higher similarity to members of group I of the E-class P450 (CYP family 548) over its entire length (Fig. S4) (Nelson, 2006). An InterProScan sequence search (http://www.ebi.ac.uk/Tools/pfa/iprscan/) (Zdobnov & Apweiler, 2001; Goujon et al., 2010) detected six regions corresponding to E-class group I signatures in Akt7-1 (Fig. S4).
To confirm the exon/intron structure, AKT7-1 cDNA was prepared from total RNA of 15A by RT-PCR using the primer pair RT7-f1/RT7-r1 (Fig. 3a, Table S2). The RT-PCR products were detected as multiple bands of 1.5–1.7 kb after gel electrophoresis (Fig. 3b). We also compared the RT-PCR products generated by 26, 28 and 30 cycles of PCR and observed similar banding patterns among the three products (Fig. 3b). These results suggested that AKT7 transcripts include multiple splice variants. The RT-PCR products generated by the 26-cycle PCR were cloned into the plasmid pGEM-T Easy, and sequences of 86 clones were obtained. Unexpectedly, the 86 clones consisted of 15 types (Type 1 to Type 15) of cDNAs derived from splice variants of the AKT7 pre-mRNA (Fig. 3c). Type 1 mRNA consists of six exon sequences with all five introns correctly spliced out and encodes the expected 522-amino acid protein. Only 18 clones (20.9%) contained Type 1 cDNA. The remaining 68 clones contained cDNAs derived from 14 aberrant splice variants of the AKT7 pre-mRNA (Type 2 to Type 15). The most frequent variant, Type 2 (34 clones), lacked 121 nucleotides at the 5′ region of the fifth exon (ΔE5), resulting from a splicing error at the 3′ end of the fourth intron. Types 3, 4, 6, 11, 12 and 13 also had the ΔE5 and single or multiple intron sequences that had not been spliced out. Type 14 contained ΔE5 and lacked 21 nucleotides at the 5′ region of the sixth exon (ΔE6) as a result of incorrect splicing at the 3′ end of the fifth intron. Type 15 also included ΔE5, ΔE6 and the second intron sequence. The other five variants (Types 5, 7, 8, 9 and 10) contained single or multiple intron sequences. The second intron escaped splicing at the highest frequency: one, 28, 11, four and six clones, respectively, included the first, second, third, fourth and fifth introns. All of the 14 aberrant splice variants were unable to encode functional proteins as a result of the presence of premature stop codons (Fig. 3c), and thus only c. 21% of AKT7 transcripts appear to be functional.
We also prepared AKT7 cDNA from total RNA of another 11 strains of the Japanese pear pathotype by RT-PCR. The RT-PCR products from all the strains showed banding patterns similar to that from 15A, indicating that the splicing error of the AKT7 pre-mRNA is a common event (Fig. 3d). We analyzed transcripts of five other AKT genes (AKT1 to AKT4 and AKTS1) with introns by RT-PCR. The cDNAs of these genes were generated by RT-PCR using gene-specific primer pairs designed to amplify full-length ORFs. The RT-PCR products were detected as single bands of the expected sizes for these genes (Fig. S5), indicating that splicing of pre-mRNAs from AKT genes other than AKT7 works correctly.
We compared the intron sizes and the sequences of the 5′ donor splice sites (six nucleotides), 3′ acceptor splice sites (five nucleotides) and putative branch sites (six nucleotides) of introns of AKT genes with the canonical sequences of fungal introns reported by Kupfer et al. (2004). Introns of the filamentous ascomycete fungi A. nidulans and Neurospora crassa fall into a narrow length range, with peak numbers of introns within a range of 50–70 nucleotides (Kupfer et al., 2004). The sizes of all five introns of AKT7-1 fall into this range, as do those of the other five AKT genes (Table S3). AKT7-1 introns tend to more frequently have mismatches to conserved sequences than other AKT genes. For example, the second intron of AKT7-1, which escaped splicing at the highest frequency, has three nucleotide mismatches in the 5′ donor splice site.
Distribution of AKT7 homologs in other pathotypes
The distribution of AKT7 homologs in 25 strains from seven pathotypes and two nonpathogenic strains of A. alternata was determined by Southern blotting (Table S1, Fig. 4). DNA from each strain was digested with HindIII, which recognizes a single site in AKT7-1, and the blot was probed with an AKT7-1 probe (Fig. 4a). The probe hybridized to two bands of the expected sizes of 1.0 and 2.4 kb in all strains of the Japanese pear pathotype (Fig. 4b). This probe also hybridized to 1.0- and 2.4-kb bands in strains of the strawberry pathotype, but not in strains of the tangerine pathotype (Fig. 4b), suggesting that AKT7-1 does not participate in EDA biosynthesis. The other pathotype strains and nonpathogenic strains did not have any homologs (Fig. 4b).
The AKT7-1 homolog of the strawberry pathotype was found to reside downstream of AFT6-1 in cosmid clone pcAFT-3 of strain NAF8 (Fig. 2). The homolog shows strong similarity to AKT7-1 (97.3% nucleotide and 96.7% amino acid sequence identity) (Figs 5a, S6). However, a nucleotide difference in the second exon causes a change of the Trp codon (TGG) in AKT7-1 to the stop codon (TGA) in the AKT7 homolog (AKT7h), indicating that AKT7h in pcAFT-3 is a pseudogene (Figs 5, S6).
AKT7 homologs of the strawberry pathotype
We previously observed that strains of the Japanese pear and strawberry pathotypes have multiple copies of the AKT and AFT genes, respectively, in their genomes (Tanaka et al., 1999; Tanaka & Tsuge, 2000; Hatta et al., 2002; Ruswandi et al., 2005). We analyzed the sequences of AKT7 and its homologs in 12 strains of the Japanese pear pathotype and 10 strains of the strawberry pathotype collected from different prefectures in Japan (Table S1).
When the entire exon and intron regions of AKT7 and the homologs were amplified from total DNA of each strain by PCR with the primer pair RT7-f1/RT7-r1 (Fig. 3a), an expected 1.9-kb DNA was generated from all strains of both pathotypes. Direct sequencing of the PCR products of each strain did not detect any single nucleotide polymorphisms (SNPs) within the products, suggesting that there are no SNPs among copies within the genome of each strain. The sequences of five strains (T88-52, T88-101, 91H-27, N18 and G90-A2) of the Japanese pear pathotype were identical to that of AKT7-1 of 15A. The sequences of the other six strains (T88-3, T88-58, Nu89-22, G16, G17 and G31) had a single nucleotide difference in the sixth exon, which causes an amino acid substitution (Glu to Arg) at position 503 (Fig. 5b). All 10 strains of the strawberry pathotype had the same sequence as AKT7h of NAF8. AKT7h had 49 nucleotide differences from AKT7-1. Of the 49 nucleotide differences, 17 cause amino acid substitutions, and one in the second exon results in a change of the Trp codon (TGG) to the stop codon (TGA) (Fig. 5b). Sequencing chromatograms for AKT7-1 of 15A and AKT7h of NAF8 and O-187 are shown in Fig. 5(c) as examples. This result strongly suggested that the strawberry pathotype strains do not have functional AKT7 orthologs.
AK-toxin production and virulence of AKT7 single-copy mutants
To determine the function of AKT7-1 in AK-toxin production, homologous recombination was employed to replace AKT7-1 with the plasmid pGDAKT7, which contains AKT7-1 interrupted with the hph cassette (Fig. S1a). Strain 15A was transformed with pGDAKT7, and the integration mode of pGDAKT7 in transformants was analyzed by PCR using primer pairs 7f3/Pt-r and Tf/7r3 (Fig. S1a). We isolated five AKT7-1 disruption transformants in which PCR produced both 1.0- and 0.9-kb fragments, showing homologous integration of pGDAKT7 into the recipient AKT7-1 locus (Fig. S1). Disruption of AKT7-1 in the transformants was also identified by Southern blotting. Total DNA of 15A and the transformants was digested with EcoRV, and the blot was hybridized with the AKT7-1 probe. The probe hybridized to an expected band of 2.4 kb in 15A (Fig. S1). The transformants had 2.4-kb bands, as did 15A, and also 2.1-kb bands, which correspond to the mutated AKT7-1 (Fig. S1). This result indicated that 15A has at last two copies of AKT7, and that these transformants have a single disrupted copy of AKT7. Four transformants also had c. 4.0-kb bands, a size consistent with the EcoRV fragment from pGDAKT7 (Fig. S1). This band might result from the unusual homologous integration of multiple copies of the vector at the AKT7 locus or ectopic integration at additional loci.
To examine the function of AKT7 in AK-toxin production, we evaluated AK-toxin production of AKT7 single-copy mutants on the basis of toxicity of their culture filtrates to leaves of Japanese pear cv Nijisseiki. Unexpectedly, culture filtrates of the mutants were apparently more toxic to pear leaves than the wild-type (Fig. 6a). The Japanese pear pathotype produces two related molecular species, AK-toxins I and II (Fig. 1), with toxin I being the more abundant and biologically active species (Nakashima et al., 1985). When AK-toxins in culture filtrates were quantified by reverse-phase HPLC, culture filtrates of the wild-type and mutant strains contained AK-toxin I, but not AK-toxin II (Fig. 7a). The single-copy mutants appeared to produce 4.0–5.4 times more AK-toxin I than the wild-type (Fig. 6b). The amount of EDA, a precursor of AK-toxin (Feng et al., 1990), in culture filtrates was also quantified by HPLC. Culture filtrates of the single-copy mutants contained 3.9–5.1 times more EDA than wild-type (Figs 6b, 7b), suggesting that AKT7 encodes a cytochrome P450 monooxygenase catalyzing a reaction resulting in reduced amounts of EDA.
The mutants were tested for virulence to Japanese pear leaves by spray-inoculation of conidial suspensions. The wild-type and mutant strains developed approximately the same number of similar-sized lesions on pear leaves with similar timing of symptom appearance (Fig. 6a), indicating that the single-copy mutants produce an excess of AK-toxin and express a similar level of virulence to the wild-type. There were no significant differences in growth rate or pigmentation on PDA, MA and CA between the wild-type and mutant strains (Fig. S7).
AK-toxin production and virulence of AKT7-1 overexpression strains
To verify the function of AKT7, we constructed AKT7-1 overexpression strains from 15A. The Type 1 cDNA of AKT7-1 was cloned between the A. nidulans trpC promoter (trpCp) and terminator (trpCt) (Mullaney et al., 1985) to make pOEAKT7 (Fig. S2a). Because the trpC promoter is constitutively active in A. alternata (Imazaki et al., 2010), we expected that it would be useful for overexpression of AKT7-1 under any conditions. The plasmid pOEAKT7 was introduced into 15A by cotransformation with the plasmid pSH75, conferring hygromycin B resistance (Kimura & Tsuge 1993), and five transformants (JpOE7-1 to JpOE7-5) were identified as carrying the trpCp-AKT7-1 cDNA-trpCt construct by PCR (Fig. S2b).
To confirm AKT7-1 overexpression in the transformants, the transcript levels of AKT7 in the transformants were analyzed in PDB cultures by real-time quantitative RT-PCR and normalized to transcripts of the actin gene (ACT) (value = 1). The transcript levels in the transformants were 2.2- to 5.2-fold higher than in wild-type (Fig. 8a). Because c. 80% of the AKT7 transcripts in the wild-type were aberrant, the functional transcript levels in the transformants were calculated to be > 10-fold higher than in wild-type.
To test for AK-toxin production, the AKT7-1 overexpression strains were grown in PDB. Culture filtrates of JpOE7-1 and JpOE7-4 showed no toxicity to pear leaves (Fig. 8b), and the amount of AK-toxin I in the culture filtrates was less than the level detectable by HPLC (Figs 7a, 8c). Culture filtrates of JpOE7-2, JpOE7-3 and JpOE7-5 showed weaker toxicity and contained smaller amounts (15.0–17.5%) of AK-toxin I than filtrates of 15A (Fig. 8c). JpOE7-4 did not produce a detectable amount of EDA, and the other transformants produced markedly less EDA than 15A (3.7–22%) (Figs 7b, 8c). The transcript levels of AKT7 in the transformants tended to negatively correlate with the amount of EDA and AK-toxin I produced (Fig. 8). These results strongly suggest that Akt7 acts to suppress AK-toxin production by converting EDA or its precursor to an unknown shunt product.
The virulence of the transformants was compared with that of the wild-type. JpOE7-1 and JpOE7-4, which did not produce any detectable AK-toxin, caused fewer lesions on leaves than the wild-type (Fig. 8d). By contrast, JpOE7-2, JpOE7-3 and JpOE7-5, which produced detectable amounts of AK-toxin I, caused lesions similar to those caused by the wild-type (Fig. 8d), suggesting that the amounts of AK-toxin produced by these three transformants are enough for expression of full virulence in laboratory tests. There were no significant differences in growth rate or pigmentation on PDA, MA or CA among the wild-type, AKT7 single-copy mutant and AKT7 overexpression strains (Fig. S7).
Toxin production and virulence of AKT7-1 overexpression strains of the strawberry pathotype
To confirm the function of Akt7-1 in limiting EDA production, we constructed AKT7-1 overexpression strains from the strawberry pathotype. The AKT7-1 overexpression vector pOEAKT7 was introduced into strain NAF8, and five transformants (SOE7-1 to SOE7-5) carrying the trpCp-AKT7-1 cDNA-trpCt construct were isolated (Fig. S2c).
The transcript levels of AKT7-1 in the transformants were tested in PDB cultures by real-time RT-PCR with the AKT7-1-specific primer pair 7re-f2/7re-r2 (Table S1). The 3′ end of each primer has a mismatch to the corresponding sequence of AKT7h of the strawberry pathotype. The primer pair did not significantly amplify AKT7h transcripts from total RNA of NAF8 (Fig. 9a). The transcript levels in the transformants were 2.4- to 5.5-fold higher than in 15A of the Japanese pear pathotype (Fig. 9a).
Culture filtrates of NAF8 showed toxicity to both strawberry and pear leaves and contained AF-toxins I and II, with toxin I being the more abundant (Figs 9b,c, S8a). Culture filtrates of SOE7-1 and SOE7-5 showed no toxicity to either strawberry or pear leaves, and the amounts of both toxins were below the level detectable by HPLC (Figs 9b,c, S8a). Culture filtrates of SOE7-2, SOE7-3 and SOE7-4 showed weaker toxicity to the leaves and contained smaller amounts of AF-toxins than the wild-type (Fig. 9c). The EDA concentration in the culture filtrates of these transformants was also markedly lower than that of the wild-type (<17.8%) (Figs 9c, S8b).
The virulence of the transformants was tested using strawberry and pear leaves. SOE7-1 and SOE7-5 caused fewer lesions on both types of plant leaves than the wild-type (Fig. 9d). By contrast, SOE7-2, SOE7-3 and SOE7-4, which produced detectable amounts of AF-toxin I, caused lesions similar to those caused by the wild-type (Fig. 9d).
We also constructed AKT7-1 overexpression transformants (TOE7-1 and TOE7-2) from strain SH20 of the tangerine pathotype, which did not have an AKT7 homolog (Fig. 4b), by introducing the vector pOEAKT7 (Fig. S2d). Introduction of the trpCp-AKT7-1 cDNA-trpCt construct decreased the amounts of EDA and ACT-toxin produced (Figs S9, S10). Together with observations of the strawberry pathotype transformants, this result confirmed that Akt7-1 participates in a shunt pathway of EDA biosynthesis.
In this study, we identified the AKT7-1 gene from strain 15A of the Japanese pear pathotype of A. alternata, which resides in the AKT cluster involved in AK-toxin biosynthesis. AKT7-1 may encode a cytochrome P450 monooxygenase. Although 15A has multiple copies of AKT7 in its genome, a single-copy mutation in AKT7 elevated production of AK-toxin and EDA, a precursor of AK-toxin. By contrast, AKT7-1 overexpression in 15A reduced production of AK-toxin and EDA. These findings show that the presumptive cytochrome P450 monooxygenase encoded by AKT7 acts to limit AK-toxin production by catalyzing a reaction converting EDA or its precursor to an unknown shunt product. Thus, the AKT cluster appears to include the gene involved in a shunt pathway of AK-toxin biosynthesis.
Host-selective toxins produced by the Japanese pear, strawberry and tangerine pathotypes have a common structural moiety, EDA, and these three pathotypes share common genes required for EDA biosynthesis. Although an AKT7 homolog (AKT7h) was identified in the strawberry pathotype strains, it appeared to be a pseudogene containing a premature stop codon. The tangerine pathotype strains did not have an AKT7 homolog. However, AKT7-1 overexpression in strains of the two pathotypes also reduced the amounts of EDA and respective toxins produced. These data confirm that Akt7 participates in a shunt pathway of EDA biosynthesis.
We found that AKT7 function is repressed by an error in splicing of the AKT7 pre-mRNA. Sequence analysis of the 86 cDNA clones obtained by RT-PCR of AKT7 transcripts identified 15 splice variants. Of the 15 types, only one (Type 1) can encode a functional protein and it accounted for only 21% (18 clones) of the 86 cDNA clones. The remaining 14 types were misspliced mRNAs, which included single or multiple intron sequences and/or incomplete exon sequences as a result of exon skipping, and are useless mRNAs that include premature stop codons. Such splicing errors of the AKT7 pre-mRNA seem to be a common event in the Japanese pear pathotype strains because multiple RT-PCR products were detected from all 12 strains tested. By contrast, we verified that the splicing of pre-mRNAs from the other five AKT genes works correctly. These results demonstrate that splicing errors are not usual in members of the AKT cluster but are unique to AKT7.
The sizes of AKT7 introns fall into the range typical of introns of the filamentous ascomycetes (Kupfer et al., 2004), as do those of the other five AKT genes examined. However, AKT7 introns tend to more frequently have mismatches to the consensus sequences of the 5′ donor and 3′ acceptor splice sites and putative branch sites of fungal introns than other AKT genes. Although frequent errors in splicing of AKT7 pre-mRNA could not be explained only by comparison of intron sequences with the fungal consensus, AKT7 certainly has a defect in the intron information content.
Alternaria alternata pathotypes have multiple copies of the toxin biosynthetic genes (Tanaka et al., 1999; Tanaka & Tsuge, 2000; Hatta et al., 2002; Ruswandi et al., 2005; Harimoto et al., 2007, 2008; Miyamoto et al., 2008, 2009, 2010; Akagi et al., 2009; Izumi et al., 2012). In experiments disrupting toxin biosynthetic genes, multiple copies of the genes were shown to be a prerequisite for the pathotypes to produce enough toxin for full virulence (Ruswandi et al., 2005; Harimoto et al., 2007, 2008; Miyamoto et al., 2009, 2010; Izumi et al., 2012). Thus, duplication of toxin biosynthetic gene clusters in the genome was the critical event for pathogens to gain significant virulence in the field. Strain 15A also had multiple copies of AKT7 in its genome. AKT7 appeared to strongly suppress AK-toxin production because a single-copy mutation caused a marked increase in AK-toxin concentration. However, there was no difference in virulence level between the wild-type and mutant strains in laboratory tests. This observation shows that the wild-type strain produces enough AK-toxin to express full virulence even though it has multiple copies of AKT7. By contrast, AKT7-1 overexpression in the wild-type strain using the trpCp-AKT7-1 Type 1 cDNA-trpCt construct caused a remarkable reduction in AK-toxin production and hence in virulence. Altogether, our results suggest that the AK-toxin production required for full virulence in the Japanese pear pathotype is regulated in a complex way by the copy number and intron information content of AKT7 and is ensured by errors in splicing the AKT7 pre-mRNA (Fig. 10). The observation that AKT7-1 overexpression in the strawberry pathotype also caused reduced virulence suggests that AF-toxin production required for full virulence is ensured by the nonsense mutation in the AKT7 homolog (Fig. 10). In the tangerine pathotype, ACT-toxin production is free from negative regulation by the AKT7 ortholog because of the absence of a homologous gene in the genome (Fig. 10).
Direct sequencing of PCR products of AKT7 and AKT7h amplified from total DNA of each strain of the Japanese pear and strawberry pathotypes, respectively, strongly suggested that there are no SNPs among copies within the genome of each strain. Recently, we determined the structure of the 1.0-Mb chromosome of strain NAF8 of the strawberry pathotype, which encodes multiple sets of the AFT cluster, and identified the entire chromosomal AFT region of c. 390 kb (R. Hatta, A. Shinjo, Y. Cho, C. Mase, C. Hara, H. Kondo, Y. Harimoto, M. Yamamoto, K. Akimitsu & T. Tsuge, unpublished results). We detected two copies of AKT7h with identical sequences in this region. Thus, the sequences of redundant copies of the gene or the homolog have been conserved after its duplication in the genome of each strain. The AKT7 sequence is highly conserved in Japanese pear pathotype strains differing in geographical origin: out of 12 strains, six had the same sequence as AKT7-1, and the other six had only a single nucleotide difference. AKT7h from 10 strains of the strawberry pathotype had identical sequences with a premature stop codon. These data suggest that the sequences of AKT7 and AKT7h of these two pathotypes were fixed at an early stage of establishment of the respective gene clusters required for biosynthesis of AK-toxin and AF-toxin and have been strictly conserved within pathogen populations. We hypothesize that such evolutionary events have been important for the pathogens to maintain significant virulence in the field.
Gardiner et al. (2009) identified two genes (FGSG_00007 and FGSG_10397) that negatively regulate both biosynthesis of deoxynivalenol (DON), a trichothecene mycotoxin, and virulence in Fusarium graminearum. FGSG_00007 and FGSG_10397 may, respectively, encode cytochrome P450 monooxygenase and a protein with a partial terpene cyclase domain, and their expression is positively regulated by the transcription factor Tri6, which regulates the TRI genes involved in the trichothecene biosynthetic pathway (Proctor et al., 1995; Gardiner et al., 2009). These genes, however, do not reside in the DON biosynthetic gene cluster. Mutations of the genes resulted in mutants with massively increased production of DON and increased virulence in wheat (Triticum aestivum). Gardiner et al. (2009) pointed out the potential of this pathogen to evolve with the ability to produce massive amounts of toxins and increased virulence. It is likely that identifying the mechanisms of negative regulation of secondary metabolism may reveal new insights into the regulation of virulence and fitness in plant pathogenic fungi.
Host-selective toxin biosynthetic genes have been isolated from six pathotypes of A. alternata (Tsuge et al., 2013).The gene clusters of A. alternata have unique features, such as the presence of multiple sets of clusters on a single chromosome, high-density distribution of transposon-like sequences in the clusters and storage of the clusters in single small chromosomes of <2.0 Mb in most strains tested (Tanaka et al., 1999; Tanaka & Tsuge, 2000; Johnson et al., 2001; Hatta et al., 2002, 2006; Masunaka et al., 2005; Ruswandi et al., 2005; Harimoto et al., 2007, 2008; Miyamoto et al., 2008, 2009, 2010; Akagi et al., 2009; Izumi et al., 2012). Loss of the small chromosomes encoding toxin biosynthetic gene clusters was observed in the strawberry, apple and tomato pathotypes, and the small chromosomes appeared to be conditionally dispensable (CD) (Covert, 1998; Johnson et al., 2001; Hatta et al., 2002; Akagi et al., 2009). The observation that most strains of the Japanese pear, tangerine and rough lemon pathotypes also carry the gene clusters on small chromosomes suggests that host-selective toxin production and hence pathogenicity of A. alternata pathotypes is controlled by these small CD chromosomes. Our data also indicate the existence of a novel mechanism of negative regulation of host-selective toxin production in A. alternata. Structural and functional analyses of CD chromosomes from different pathotypes may provide insight into the origin of the CD chromosomes controlling host-selective toxin production and the evolution of toxin production and pathogenicity in A. alternata pathotypes.
We thank Yasuki Tahara for providing and maintaining plant material, Kosuke Hanada for valuable suggestions and Norio Nakazawa, Fusaharu Nakatani, Yutaka Sato and Motoaki Kusaba for providing fungal strains. Most of this work was supported by Grants-in-Aids for Scientific Research (A) (23248007 to T.T.) and Scientific Research (S) (19108001 to T.T. and 21228001 to K.A.) from the Japanese Society for Promotion of Sciences and Special Coordination Funds for Promoting Sciences from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (T.T.).