Invasions of gladiolus rust in North America are caused by a widely-distributed clone of Uromycestransversalis

Isolate collection and DNA extraction

DNA was extracted from a total of 92 samples of Uromyces transversalis (Table 1) in preserved leaf tissue obtained from Australia (n = 7), New Zealand (n = 10), Mexico (n = 60), or as fresh, infected leaf tissue collected within the USA (n = 10) or from border interceptions from Costa Rica (n = 5). The preserved tissue from Australia and New Zealand had been stored in envelopes in herbaria, whereas the tissue from Mexico had been stored individually in conical polypropylene tubes at −20 °C. From each preserved leaf tissue sample, DNA was extracted using a modified genomic DNA mini-prep protocol (Lee, Milgroom & Taylor, 1988). Briefly, multiple uredinia were scraped to remove urediniospores (0.02–0.04 g) using a sterilized scalpel and transferred into 1.5 mL reaction tubes containing 246.9 µL lysis buffer (150 µL sddH₂O, 25 µL 0.5 M EDTA (pH 8), 25 µL 1.0 M Tris, 43.75 µL 20% SDS solution, 3.15 µL 20 mg/L proteinase K and 0.0025 g NaHSO₃) (Thermo Fisher Scientific, Watham, MA). Sample tubes were vortexed for 1 min, incubated at 65 °C for 15 min, and centrifuged (13,978× g for 5 min). The precipitates were discarded and the supernatants transferred to new 1.5 µL microcentrifuge tubes. A solution of 50 µL 7.5 M NH₄OAc (Sigma-Aldrich, St. Louis, MO) was added to each tube, vortexed for 10 s, and tubes were chilled on ice for 15 min. Samples were then centrifuged (13,978× g for 3 min). The supernatants were again transferred to new 1.5 mL microcentrifuge tubes and 175 µL isopropanol was added, tubes were mixed and centrifuged (13,978× g for 5 min), and the supernatant was discarded. The pellets were rinsed twice with 250 µL of 70% ethanol solution, dried and re-suspended with 25 µL sddH₂O, then incubated at 30 °C for 10 min. All samples were stored at −20 °C until further use. DNA extractions from fresh, infected leaf tissue was performed using a modified hexadecyltrimethylammonium bromide (CTAB) protocol (Rodgers & Bendich, 1985). Briefly, one to three one cm² excised pieces of infected leaf tissue were frozen and using a mortar and pestle, ground to a fine powder in liquid nitrogen. DNA extraction buffer (100 mM Tris, pH 7.5; 1% CTAB; 0.7 M NaCl; 10 mM EDTA; 1% 2-mercaptoethanol; 0.3 mg/ml proteinase K) was added to the ground tissue and incubated at 65 °C for 30 min, followed by two rounds of chloroform: isoamyl alcohol (1:1) extraction, and precipitated with 2-propanol. DNA was resuspended in TE buffer containing 1 mg/ml RNase.

Three U. transversalis isolates collected in the USA were maintained at the USDA Agricultural Research Service, Foreign Disease-Weed Science Research Unit, biosafety level-3 plant disease containment facility at Ft. Detrick, Maryland. These samples were propagated from U. transversalis- infected Gladiolus plants collected from California and Florida commercial fields in 2011 and 2014 (Table 1). Prior to extraction using the modified CTAB protocol described above, urediniospores were harvested from infected Gladiolus plants with a microcyclone spore collector (Cherry & Peet, 1966; Peterson & Berner, 2009; Tervet et al., 1951). Spores were germinated by placing 300 mg of freshly harvested spores in a 23 cm × 33 cm glass container that contained 300 mL of sterile water with 15 µg ampicillin. A sterile wooden applicator stick was used to break up clumps of spores, so that the spores were evenly distributed across the surface of the water. The container was covered and left in the dark overnight (16–18 h). The germinated spores were then scraped from the surface of the water, blotted dry with sterile-paper towels and stored in −20 °C.

Genome sequencing and assembly

Genomic DNA obtained from three isolates (CA11-1, FL11-1, and FL11-2) was standardized to 50.0 ng/µL using a nanodrop and sent to the Georgia Genomics Facility (GGF) (University of Georgia, Athens, GA) for library preparation and sequencing using the Illumina MiSeq platform as 300-bp paired ends reads using a 600 cycle cartridge with a NGS library preparation method. The raw forward and reverse reads of each isolate was observed using FASTQC v.11.2 (Babraham Bioinformatics Institute). Quality control was performed using FASTX-Toolkit v.3.0.13 (http://hannonlab.cshl.edu/fastx_toolkit/). All reads with a phred score below Q = 22 were removed prior to assembly. ABySS v.1.3.6 (Simpson et al., 2009) was used for de novo assembly of forward and reverse reads into contiguous sequences (contigs) for each isolate, using an optimal K-mer value of 64 determined with multiple assembly trials. Generated contigs were then imported into Geneious v.6.1.8 (Kearse et al., 2012).

Microsatellite discovery and marker development

To increase the potential for successful microsatellite marker development, only contigs 200 bp or greater in size with matched pair reads were considered. Contigs and singletons were searched for at least five perfect repeats of trimeric, tetrameric, pentameric, and hexameric motifs in Geneious v.6.1.8 using Phobos v.3.3.12 (Kearse et al., 2012; Mayer, 2006). Mono and dinucleotide repeats were eliminated due to the difficulty of scoring allele differences. Contigs with microsatellites identified using our criteria were aligned by multiple sequence alignment using Geneious Align v.6.1.8. default parameters. Microsatellites shared among the three isolates were visually assessed for sequence variation. Those that showed microsatellite repeat number variation among isolates and had at least 50 bp flanking each side of the repeat were considered acceptable for primer design. Primers for amplification of microsatellite loci were designed with Primer3web v.4.0 (Koressaar & Remm, 2007; Rozen & Skaletsky, 1999; Untergasser et al., 2012) to produce amplicons of approximately 180–350 bp in length with an optimal annealing temperature of 59 °C.

Sixty primer pairs for candidate microsatellite loci were initially evaluated on the three sequenced isolates CA11-1, FL11-1, and FL11-2 to verify that the PCR worked with the designed primers and that the PCR amplicons were the expected size. PCR was carried out in 10 µL reactions with 1 µL of 10 × ExTaq buffer (Takara Bio Inc., Mountain View, CA), 1 µL of 2.5 mM dNTPs (Takara Bio Inc.), 0.25 µL of 10 µM forward primer, 0.25 µL of 10 µM reverse primer (Integrated DNA Technologies, Coralville, IA), 0.1 µL of ExTaq polymerase (Takara Bio Inc.), 6.9 µL of sterile distilled H₂0, and 0.5 µL of 50.0 ng/µL DNA template. Reaction conditions were 94 °C for 2 min followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 59 °C for 30 s and extension at 72 °C for 30 s, followed by a final extension of 72 °C for 5 min. Amplification of PCR products within the expected size range was confirmed by electrophoresis run at 95 V (4.75 V/cm) on a 2% (wt/vol) agarose gel (Alfa Aesar, Haver Hill, MA) for 2.5 h using a 100 bp size standard (New England Biolabs Inc., Ipswich, MA).

Twenty-five primer sets (Table 2) that successfully amplified the three sequenced isolates were screened for polymorphism on a panel of U. transversalis that included seven additional isolates from California (Table 1). A three-primer method (Schuelke, 2000) was used in this round of marker evaluation. The forward primer for each candidate marker had a CAG tag (5′-CAGTCGGGCGTCATCA-3′) (Hauswaldt & Glenn, 2003) added to the 5′ end. The third primer consisted of the CAG tag, labeled with a 6FAM fluorescent dye (Invitrogen Inc., Carlsbad, CA) on the 5′ end. PCR was carried out in 12 µL reactions with 1.2 µL of 10 × ExTaq buffer (Takara Bio Inc., Mountain View, CA), 1.2 µL of 2.5 mM dNTPs (Takara Bio Inc.), 0.1 µL of 10 µM forward primer, 0.5 µL of 10 µM reverse primer (Integrated DNA Technologies, Coralville, IA), 0.5 µL of 10 µM 5′6FAM-labeled CAG tag primer (Invitrogen Inc.), 0.1 µL of ExTaq polymerase (Takara Bio Inc.), 7.9 µL of sterile distilled H₂0, and 0.5 µL of approximately 50.0 ng/µL DNA template. Reaction conditions were 94 °C for 2 min followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 30 s, followed by a final extension of 72 °C for 5 min. Amplification of individual PCR products within the expected size range was confirmed by electrophoresis.

One microliter of a 1:10 dilution of PCR product was added to 0.1 µL of GeneScan 500 LIZ-labeled size standard and 9.9 µL of Hi-Di formamide (Applied Biosystems Inc., Foster City, CA). Amplicons were denatured by incubation at 95 °C for 5 min and immediately placed on ice. Fragment analysis was conducted at the GGF on an Applied Biosystems 3730xl 96-capillary DNA Analyzer. GeneMapper v.4.0 (Applied Biosystems Inc.) and Geneious v.6.1.8 (Kearse et al., 2012) were used to determine allele sizes from the chromatograms.

Multiplex PCR

Eight primer sets (Table 2, see loci with asterisks) that consistently produced fragments within the expected size range for the 10 isolates were optimized for multiplex PCR. Two multiplex reactions (multiplex 1 –Ut513, UtCA759, Ut2648 and Ut3161; multiplex 2 –Ut497, Ut1841, Ut1908 and Ut2048) were developed to increase efficiency and decrease cost for genotyping a large panel of isolates. The forward primers of the microsatellite markers selected for multiplex PCR were labeled at the 5′ end with one of the fluorescent dyes from the DS-33 dye set: 6-FAM (Integrated DNA Technologies), VIC, PET, or NED (Applied Biosystems Inc.). Multiplex reactions were optimized so that loci with alleles of similar size ranges were labeled with different dyes. All 92 samples were genotyped with the eight markers in the multiplex reactions.

Multiplex PCR was conducted using a modified protocol of the Type-it Microsatellite PCR kit (Qiagen, Hilden, Germany) in 10 µL reactions with 5 µL of 2 × Type-it Master Mix buffer, 1 µL of 10 × primer mix (2 µM of each primer in the multiplex), 3 µL of sterile distilled H₂0, and 1 µL of approximately 50.0 ng/µL DNA template. Reaction conditions were 94 °C for 2 min followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 30 s, followed by a final extension of 72 °C for 5 min. Amplification of PCR products within the expected size range was confirmed by electrophoresis. The PCR products were prepared as described above and sent to GGF for fragment analysis. Geneious v.6.1.8 (Kearse et al., 2012) was used to determine allele sizes from the chromatograms. Loci were distinguished by fluorescent dye. Only peaks above the relative intensity cutoff threshold of 500 relative fluorescence units (RFU) were scored.

Whole genome sequencing and assembly

The Illumina MiSeq PE 300 sequencing platform generated 32,461,282 reads with an average insert size of 575 bp and read lengths of 301 bp. Sequence quality was assessed by phred score and signal purity filter values resulting in a total of 25,452,493 reads with a P F of 99.26%, which corresponds to 6.02 to 9.98 million reads for CA11-1, FL11-1, and FL11-2 (Table 3).

The de novo draft assemblies resulted in 5,706,372 (94.7% assembled out of the total filtered reads), 4,305,978 (43.2%), and 7,444,849 (80.4%) total assembled reads for CA11, FL11-1, and FL11-2, respectively. This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accessions PTJR00000000, PTJQ00000000, and PTJP00000000 for CA11, FL11-1, and FL11-2, respectively. Using Geneious v.6.1.8 (Kearse et al., 2012), contigs for each isolate were filtered to select only those 200 bp in length or greater. This resulted in 466,181, 548,017, and 645,533 contigs for CA11, FL11-1, and FL11-2, respectively, which were subsequently used to search for microsatellite repeats (Table 3).

Microsatellite marker development

An alignment of the 18,950 contigs produced 4,296 contigs with potentially informative microsatellite loci shared among the three isolates. Microsatellite loci shared by at least two of the three isolates were observed in 2,754 of the aligned contigs. Microsatellites were identified in 0.98%, 1.03%, and 1.34% of the contigs, showing that the discovery rate of microsatellites was consistent among sequenced isolates.

Sixty sets of primers were developed and screened by PCR on the three isolates of U. transversalis. Of the 60 putative markers, 25 were successfully amplified by PCR and evaluated for polymorphism on the panel of ten isolates from the United States (Tables 1, 2). To determine if contigs used for marker development were sequences of U. transversalis and not contaminant sequence, we used blastn at NCBI (Altschul et al., 1990) to identify the sequences in the database that were most similar with high coverage. Most often the top hits were Puccinia graminis, another rust species, or another fungus, with percent identity always <90%. Sometimes the top hit was another eukaryote, but it was an unlikely contaminant, such as mouse, sheep or fish. The percent identity for these was always <90% and the e-values were usually high (>0.01) or the query coverage was low (<10%). We concluded that the contigs selected for marker development were sequences of U. transversalis and not contaminants. Isolate CA14-7 consistently produced peaks below our cutoff of 500 RFU for reliable data; however, there were fragments at the expected allele size ranges. It is possible that there were PCR inhibitors in the DNA extract or that the DNA was of lower quality than the other samples. Of the 25 markers, seven were monomorphic, with only one allele each, and 18 were polymorphic, with two alleles each (Table 2). All 10 isolates from the United States were the same genotype based on the 25 markers. The polymorphism was identified among alleles within each locus, rather than among individuals. Overall, the microsatellite markers showed allelic diversity, but no genotypic diversity among the isolates from the United States. There was a high heterozygosity within individuals with each isolate having both alleles for the polymorphic loci (Table 2).

Genotypic analyses

Eight primer sets that consistently produced fragments within the expected size range for the 10 isolates were optimized for multiplex PCR. We selected markers that were polymorphic and did not produce alleles in overlapping size ranges. Unfortunately, after subsequent genotyping runs one of the markers UtCA759 was determined to be monomorphic. Even though the four markers within each multiplex were run with different fluorescent dyes, we minimized size overlap to prevent pull-up effects or bleed through from the fragments run with the other dyes. When using the eight microsatellite markers in two multiplex reactions, samples from the USA, Costa Rica and Mexico consistently produced PCR products of the expected size. Samples from New Zealand and Australia produced inconsistently sized PCR products despite duplicate reactions. Using the same eight markers, we attempted to genotype 16 leaf samples from South Africa, but these repeatedly failed to produce PCR products. Only one isolate, PREM 57128, which was sampled in 1998 produced a faint PCR product; however, no fragments were detected in the analysis.

Fragment analysis showed allelic variation within individuals, but no genotypic variation was observed among the isolates from Australia, Costa Rica, Mexico, New Zealand, and the USA. In all cases where fragments were observed and were above the relative intensity cutoff of 500 RFU, the genotypes were identical to each other and to all isolates from the USA (Table 2). In some cases where the peaks were below the relative intensity cutoff of 500 RFU, there was a fragment for only one of the alleles or the alleles were a slightly different size; however, these results were not reliable (Table S1). Marker Ut497 consistently failed to produce fragments or peaks above the relative intensity cutoff for almost all isolates where DNA was obtained from preserved leaf tissue. The six remaining markers (Ut513, Ut1841, Ut1908, Ut2048, Ut2648 and Ut3161) were polymorphic with only two allele sizes observed for each marker, while one marker (UtCA759) was monomorphic, producing only one allele size.

Sequence divergence

Visual assessment of the aligned contigs for the three sequenced isolates revealed that there were two distinct haplotypes (nucleotide sequence patterns) for each isolate occurring at nearly all microsatellite loci (Table 4, Fig. 2). The variation in repeat number occurred between alleles or haplotypes of the same isolate. Additionally, there were numerous single nucleotide polymorphisms (SNPs) detected in the regions flanking the microsatellite repeats. The two haplotypes within each of the three isolates sequenced were identical to the haplotypes among all three isolates for the loci compared, including microsatellite loci that were monomorphic based on sequence length. Variation in the microsatellite-flanking sequences between haplotypes was estimated for loci with complete data sets for the three isolates. The two alleles differed in nucleotide sequence by 1.6% to 6.9% (Table 4).