Development of Genomic Resources for the Powdery Mildew, Erysiphe pulchra

Powdery mildews (PMs) (Ascomycota: Erysiphales: Erysiphaceae) are widespread plant pathogenic fungi affecting over 10,000 angiosperm species including many economically and agriculturally important plants (Glawe 2008; Spanu et al. 2010). The PMs are a single evolutionary lineage with a taxonomically diverse set of species belonging to five tribes: Blumerieae, Erysiphaceae, Cystotheceae, Golovinomyces, and Phyllactinieae (Takamatsu 2013). Plant-fungus interaction is initiated with the contact between the host and spore surfaces (Łaźniewska et al. 2012). Almost all PMs have ascocarp appendages that vary among genera/species and have been used since 1851 (Léveillé 1851) to delineate between different PMs (Takamatsu 2013). The use of morphology to phylogenetically delimit PM species quickly breaks down and scientists have turned to molecular sequencing to resolve evolutionary relationships among PMs. Only a subset of PMs have undergone ITS or rDNA sequencing, and even with those sequencing efforts, there is a critical need for multigene sequencing to further resolve closely related lineages (Meeboon and Takamatsu 2017; Takamatsu et al. 2015). To date, the genomes of only four PM species have been sequenced. Spanu et al. (2010) generated a reference genome for Blumeria graminis f. sp. hordei (host Hordeum vulgare, Blumerieae) and low coverage draft genomes for Erysiphe pisi and Golovinomyces orontii. Investigators at the Max Planck Institute have undertaken additional genomic sequencing of the E. pisi and G. orontii PM species (https://www.mpipz.mpg.de/powdery_mildew_project_description), and the sequence data are publicly available, but a publication is still being prepared. Jones et al. (2014) sequenced a number of isolates of E. necator, the PM that is a major worldwide pathogen of grapes. These genomes are large (>120 Mb) compared with other fungi, are full of retrotransposons, and encode species specific candidate secreted effector proteins (CSEPs).

Often overlooked are the negative aspects of PM in the ornamental industry where symptoms decrease aesthetic value due to white and brown discolorations on leaves, misshapen leaves, malformed or lack of flowers, and overall decreased plant size. Lack of sales, loss of plants, and cost of control, usually through fungicide spraying, all contribute to major economic losses across the ornamental and horticultural industry. E. pulchra is the causal agent of PM on large bracted dogwood species, Cornus spp. In 1994, the disease reached epiphytotic levels in the United States. Tens of millions of dollars’ worth of flowering dogwoods (C. florida) were destroyed and millions of cultivated seedlings lost their commercial value because EPA-approved disease management strategies were not established (Li et al. 2009). Distribution of E. pulchra has been documented in Asia and Europe (Bai et al. 2014; Farr and Rossman 2017; Garibaldi et al. 2009), leading to severe outbreaks of flowering dogwood in several gardens and nurseries in northern Italy (Garibaldi et al. 2009).

Currently, genomic and genetic resources available to understand or combat the damage caused by PMs are lacking and phylogenies are largely unresolved. There is a critical need to expand the scope of research to additional PM pathosystems and describe the molecular and physical mechanisms of plant-microbe interactions that directly and negatively impact U.S. agriculture, particularly the ornamental industry. Therefore, genome sequencing of E. pulchra was undertaken to develop a draft genome that can be used for host/pathogen studies and for developing molecular markers for population studies.

Genomic DNA of E. pulchra (TENN-F-071826) was isolated from infected C. florida leaves located at the University of Tennessee, Knoxville, TN. To isolate pure PM DNA, conidia were lightly brushed from the leaf surface into a sterile Petri dish and DNA isolated with the DNeasy Plant Mini Kit (Qiagen, U.S.A.). As an additional control to ensure that only PM DNA was isolated for genome sequencing, we isolated DNA from two other samples (PM-infested leaf tissue and noninfested leaf tissue). The ITS 18S rRNA subunit region from three samples of DNA was amplified and sequenced (White et al. 1990). When the PCR products were visualized on a 2% agarose gel stained with ethidium bromide, a single amplicon of ∼600 bp was present for the PM sample of conidia, a single amplicon of ∼700 bp was present for the noninfested leaf tissue sample, and two amplicons (∼600 bp and ∼700 bp) were amplified for the coisolation of DNA from conidia and plant. BLAST of the resulting sequences identified the ∼600 bp amplicon as E. pulchra and the ∼700 bp amplicon as C. florida. The two amplicons in the fungus/plant ITS coamplification were separated on 10% denaturing acrylamide gels and the bands visualized with a silver stain. The individual bands were excised from the gel, used as DNA templates for another PCR, purified, and resequenced (Trigiano and Ownley 2017; Trigiano et al. 2016). The individual ITS amplicons were identified as E. pulchra and C. florida. This indicated that only PM DNA was isolated and was used for library construction.

A genomic DNA library (400 bp single-end read) was developed for E. pulchra by following the manufacturer’s instructions using the Ion Xpress Plus gDNA Fragment Library Preparation kit (Life Technologies, Carlsbad, CA). After fragment preparation and prior to sequencing, the library was quantified using the Ion Library Quantitation Kit (Life Technologies) to determine the appropriate library dilution. Following library dilution, the library was enriched with Ion Sphere Particles using the Ion PGM Template OT2 400 Kit (Life Technologies) in preparation for sequencing. The library was sequenced on the Ion PGM System (Life Technologies) using the Ion PGM Sequencing 400 Kit (Life Technologies) and Ion 318 Chip v2 (Life Technologies). Three sequencing runs were used for adequate coverage, totaling 12.6 million reads with 319 bp average read length totaling 4.0 Gb of reads. The genome assembly was done using SPAdes version 3.10.1 (Bankevich et al. 2012) with kmer sizes of 21, 33, 55, 77, 99 and the “careful” option. BUSCO version 3.0.2 was used to train Augustus (Simão et al. 2015) using the BUSCO fungi_odb9 dataset for the lineage option. BUSCO reported 99.0% complete BUSCOs (C:99.0% [S:98.3%, D:0.7%], F:0.3%, M:0.7%, n: 290). Maker (Cantarel et al. 2008) was then used to integrate ab initio gene predictions from Augustus (Stanke and Morgenstern 2005) and GeneMark (Besemer and Borodovsky 2005) with protein homology evidence from the Swiss-Prot protein database (ftp://ftp.uniprot.org/pub/databases/uniprot/​current_release/knowledgebase/complete/uniprot_sprot.fasta.gz, accessed 9/29/2017) and the protein sequences from several E. necator (GenBank accession no. ASM79871v1) and B. graminis (GenBank accession no. ASM15106v3). BLASTP was used to search for CSEPs in E. necator. The cutoff for identification of CSEPs in E. pulchra were percent identity (pident) of >30%, query coverage (qcovs) of >50%, and e-value <1e–10. MISA (http://pgrc.ipk-gatersleben.de/misa/) was used to identify microsatellites within the genome assembly, with a minimum number of 10 repeats for mononucleotides, six repeats for dinucleotides, and five repeats for tri, tetra-, penta-, and hexanuleotides.

The draft genome assembly of E. pulchra was 63.5 Mbp and resulted in formation of 19,442 contigs (N50 = 11,686 bp) that resulted in an estimated genome coverage of 62×. Gene annotation predicted a total of 6,860 genes, with 6,665 genes with over 100 amino acids in length. We found 102 of the CSEPs reported for E. necator (Jones et al. 2014) to be present in E. pulchra. A total of 18,244 microsatellites were found in the draft genome. The most numerous repeats were mononucleotide repeats (15,081). Di- (1,233), tri- (1,090), and tetranucleotide (614) repeats were the next most common motifs followed by the penta- (107) and hexanucleotide (119) motifs. We selected 18 microsatellite loci by searching 500 randomly selected contigs. Screening of the 18 loci against four E. pulchra samples, one each from Maryland, Mississippi, New Jersey, and Tennessee, revealed that nine loci were suitable for further characterization (GenBank accessions MG516572 to MG516580). The nine loci were used to genotype DNA isolated from PM infected leaves that were collected in Maryland (n = 3), Mississippi (n = 15), New Jersey (n = 12), and Tennessee (n = 17). A total of 14 alleles were detected and five loci were monomorphic, whereas four loci were polymorphic. Shannon’s information index was 0.12 and ranged from 0.00 to 0.45. Overall diversity was 0.07 and ranged from 0.00 to 0.40. Private alleles were detected in the Mississippi and Tennessee samples. Clustering of the samples using two-dimensional principal coordinates analysis explained 65% of the variation and the samples clustered independent of geographical location.

We expect the draft genome of E. pulchra will prove a valuable resource in further understanding of the evolutionary history of the PMs and will also provide insight into strategies that led to the wide host expansion and environmental adaptations so effectively employed by the PM lineages. Lastly, the microsatellites identified and validated from the draft genome sequences will be useful in future research to determine pathogen population structure and patterns of dissemination.

This whole-genome shotgun project has been deposited at DDBJ/ENA/GenBank under the accession number PEDP00000000. The version described in this paper is version PEDP01000000.