Background

Intraspecific genomic comparisons between pathogenic and non-pathogenic fungal isolates can elucidate key factors required for pathogenicity in plant hosts [1, 2]. These factors include unique genes that are not only vital for the pathogen to cause disease, but can also provide tools to differentiate pathogens and non-pathogens for detection and monitoring. Variation in genome size, predicted effectors, and unique genome regions have been used previously to differentiate plant pathogens from saprophytic non-pathogens [3,4,5]. Just as the hosts exhibit variation in resistance, fungal pathogens (e.g., Fusarium oxysporum) also exhibit variation in virulence [6]. Previous analyses of F. oxysporum f. sp. lycopersici and F. oxysporum f. sp. radicis-cucumerinum chromosomes identified conditionally dispensable chromosomes containing virulence factors that could be transferred between compatible strains through heterokaryon formation [4, 7]. Other previous studies found that these virulence factors were encoded by diverse genes. Fusarium transcription factors and their corresponding virulence genes, such as those encoding proteins secreted in xylem (SIX), were found to be essential for pathogenesis and high virulence in most F. oxysporum formae species [8,9,10,11]. Based on carbohydrate-active enzyme analysis, the copy number and types of plant cell wall-degrading enzymes have been found to be important for pathogenesis of F. oxysporum formae species in legumes [12]. In addition, phytotoxic secondary metabolites and associated transporters (i.e., fusaric acid, fumonisin B1, and beauvericin) have been identified as important for pathogenesis in the F. oxysporum f. sp. cubense [12, 13], and transposable elements (i.e., miniature impala elements) have been documented to enhance effector diversity [14,15,16]. These factors help distinguish pathogenicity and/or virulence in Fusaria and could also provide a basis for early detection of pathogens which would facilitate disease management programs through efforts to reduce pathogen spread to uninfested sites.

Fusarium oxysporum is a cosmopolitan, ascomycete fungus with observed functions that range from an innocuous soil saprophyte to a catastrophic plant pathogen in important agricultural crops and ecologically important tree species [17]. In addition to these diverse functions, F. oxysporum is also known to display high levels of host-specificity. These highly host-specific pathogenic strains are classified into formae speciales (forma specialis singular) that exhibit high genomic plasticity, which has been attributed to effector profiles housed on one to many conditionally dispensable, lineage-specific accessory chromosomes, where the vast majority of genomic variation exists, in comparison to a set of shared “core” chromosomes containing essential genes for normal cellular function [16, 18,19,20]. These lineage-specific chromosome(s) house genes involved in virulence and/or evasion of plant host detection and have been described as repeat- and transposon-rich [1, 3]. Comparisons of these virulence genes through whole-genome analyses are useful for distinguishing pathogens from non-pathogens on the susceptible host of a specific F. oxysporum forma specialis [1]. In this study, we analyzed and compared the genomes of pathogenic and non-pathogenic F. oxysporum strains, which were both isolated from the roots of Acacia koa A. Gray (koa).

Gardner (1980) described Fusarium wilt disease in low- to mid-elevation, below 610 m (2000 ft), forests in the Hawaiian archipelago affecting koa, an endemic tree species [21, 22]. Koa trees are ecologically, economically, and culturally valuable to Hawai‘i. Koa is the second most abundant tree in Hawai‘i’s forests and many of Hawai‘i’s indigenous flora and fauna rely on koa for food and shelter [21]. Due to koa’s importance to Hawai‘i, efforts have been increased to reincorporate koa into lowland forests, where its populations have declined [23]. Currently, koa wilt disease [caused by F. oxysporum f. sp. koae (Fo koae)] represents the predominant threat to koa across its native range [24].

Management methods for koa wilt disease include fumigation and pesticides that are environmentally harmful and economically costly when used in forests and/or timber stands. As a result, current efforts are focused primarily on producing disease-resistant seedlings derived from seed collected from natural stands [23]. To find naturally occurring resistance to the pathogenic strains of the fungus (Fo koae), wild-collected seed is used for extensive disease resistance-screening assays in the greenhouse [25]. These screening trials are costly and require diverse pathogenic Fo koae strains to ensure that the disease resistance is robust in the selected germplasm. Consequently, these disease resistance-screening programs would greatly benefit from faster, more cost-effective strategies for screening pathogenic Fo koae isolates. This screening process is further complicated by considerations that pathogenic fungi can evolve quickly; much faster than their plant hosts [1, 26, 27]. For this reason, new, genetically diverse strains of the pathogen must be continually incorporated in screening programs to help ensure that disease resistance remains stable against newly evolved or introduced strains of the pathogen in Hawai‘i’s forest and agricultural lands. Collecting, isolating, and characterizing pathogenic Fo koae isolates is an arduous and time-consuming process that requires 3-month greenhouse virulence assays to confirm pathogenicity. Pathogen-specific, PCR primers would allow for rapid detection of pathogenic Fo koae strains, which would expedite collections of new pathogenic strains for disease resistance-breeding programs to meet the needs of growers and land managers for restoration and/or timber purposes.

Objectives for this study were to 1) conduct and analyze whole genome sequences of one highly virulent Fo koae isolate and one non-pathogenic F. oxysporum (Fo) isolate both collected from koa; 2) identify virulence-associated genes with putative roles in koa wilt disease; 3) use the identified differences to develop Fo koae-specific PCR primers; and 4) test the robustness of the developed primers to identify highly virulent Fo koae isolates among uncharacterized F. oxysporum populations.

Results

Whole genome sequencing and assembly

To conduct whole genome comparisons, one pathogenic Fo koae isolate (44) and one non-pathogenic Fo isolate (170) were whole genome sequenced using Illumina TruSeq short-read sequencing (150 bp). Libraries resulted in 397,390,528 reads with 1245x coverage for pathogenic isolate Fo koae 44 and 385,805,256 reads with 1150x coverage for non-pathogenic isolate Fo 170. De novo assembly of both Fo koae 44 and Fo 170 identified genome sizes of 48 and 50 Mb, respectfully. The QUAST report metrics are shown in Table 1.

Table 1 QUAST assembly statistics of the SPAdes de novo genome assembly of the pathogenic Fusarium oxysporum f. sp. koae 44 (Fo koae 44) and non-pathogenic F. oxysporum 170 (Fo 170) isolates
Full size table

Whole genome phylogenetic analysis

A whole-genome maximum likelihood phylogeny was constructed to elucidate the evolutionary relationships of Fo koae 44 and Fo 170 compared with other Fusarium spp. and F. oxysporum formae speciales (Additional File 9). The phylogeny showed that both Fo koae 44 and Fo 170 grouped into a well-supported [Bootstrap (BS) = 100] clade with other F. oxysporum formae speciales (Additional File 1). Interestingly, both isolates grouped into separate sub-clades with distinct F. oxysporum formae speciales, rather than grouping together (Additional File 1). Fo koae 44 clustered in a sub-clade (BS =100) that included F. oxysporum f. sp. cubense race 1 and Fo 170 clustered in a sub-clade (BS =100) with F. oxysporum f. sp. gladioli. This result suggests that these isolates were indeed F. oxysporum and that Fo koae 44 and Fo 170 were sufficiently, genetically different to allow comparison for identifying lineage-specific DNA necessary for pathogenicity in koa.

Genomic analyses of Fo koae 44 and Fo 170

Genomic sequences of Fo koae 44 and Fo 170 were aligned with reference genome sequences of a well-characterized F. oxysporum f. sp. lycopersici isolate (GenBank accession AAXH00000000.1) to identify regions of highest dissimilarity and identify lineage-specific DNA, with potential use for pathogen-specific primer development. Based on the reference genome, 11 putative “core” chromosomes shared synteny with Fo koae 44 and Fo 170. Putative lineage-specific DNA (LSX) was identified in each genome that did not map to the genome of F. oxysporum f. sp. lycopersici or show synteny with each other (Fo koae 44 and Fo 170) or genomes of other F. oxysporum formae speciales (Fig. 1a). Low-coverage sequences and non-Fusarium DNA were discarded to remove possible DNA contamination. The average nucleotide identity (ANIm) between Fo koae 44 and Fo 170 was 96.5%. The remaining 3.5% genetic dissimilarity is largely due to the identified LSX. Further analyses comparing the putative chromosomes of Fo koae 44 and Fo 170 indicated that variants (SNPs and indels) were localized on specific chromosomes (LSX_44 and LSX_170) and at chromosome ends (Fig. 1b). Putative core chromosomes 1, 2, 4, 5, 7, 8, 9, and 10 were the most similar with identified variants localized on chromosome ends, likely due to higher recombination observed at telomeres or assembly errors [20, 28]. Putative chromosomes 11, 12, 13, and LSX had the most variation (Fig. 1b). Analyses of the distribution of repeats and transposons determined that F. oxysporum transposable elements (e.g., Foxy, Skippy, Impala, Hop, Mariner, Hat, and Helitron) occurred on the LSX of both Fo koae 44 and Fo 170 [29]. The average core chromosome was comprised of 3 and 4% transposable elements and repeats in Fo koae 44 and Fo 170, respectively. In contrast, the transposable elements and repeats comprised 19 and 20% of LSX for Fo koae 44 and Fo 170, respectively (Fig. 2a & b). This localization of transposable elements provides evidence that differences within the LSX are likely attributable to one or more conditionally dispensable chromosome(s) in both Fo koae 44 and Fo 170.

Fig. 1
figure 1

a Synteny map showing synteny between each of 11 putative core chromosomes that were mapped to Fusarium oxysporum f. sp. lycopersici (reference strain). Most of the putative twelfth chromosome (putative lineage-specific DNA) denoted as LSX_44 for F. oxysporum f. sp. koae 44 (Fo koae 44) and LSX_170 for F. oxysporum 170 (Fo 170) does not show synteny between the pathogen and non-pathogen and did not map to the reference strain. This lack of synteny on the putative twelfth chromosome is indicative of lineage-specific DNA, putatively containing conditionally dispensable, accessory chromosomes. b Distribution of variants (single nucleotide polymorphisms, SNPs) comparing the putative chromosomes of the koa wilt pathogen Fo koae 44 and non-pathogenic Fo 170. Darker bars indicate more SNPs identified in that region. Putative chromosomes 11, 12, 13, and putative lineage-specific DNA (ChromosomeX; LSX) show the highest SNP densities

Full size image
Fig. 2
figure 2

a Distribution of transposons (orange bars) and repeats (purple bars) across the putative chromosomes and putative lineage-specific DNA (LSX) of Fusarium oxysporum f. sp. koae (Fo koae 44; koa wilt pathogen). b Distribution of transposons (orange bars) and repeats (purple bars) across the putative chromosomes LSX in Fusarium oxysporum (Fo 170). Majority of the transposons were identified in the LSX

Full size image

Genome annotation

Using the Maker pipeline, 15,380 and 15,763 transcripts and corresponding proteins were predicted for Fo koae 44 and Fo 170, respectively, for use in identifying genes associated with pathogenicity in koa wilt disease. Using OrthoVenn2, 1760 non-orthologous proteins from 78 gene clusters and 1576 single genes, were identified as exclusive to Fo koae 44 when compared to Fo 170. In comparison, Fo 170 had 2123 exclusive non-orthologous proteins from 137 gene clusters and 1780 single genes. Fo koae 44 had enrichment of gene ontology (GO) terms for nucleic acid binding (Additional File 2), whereas Fo 170 had enrichment of GO terms for GTP binding, regulation of transcription from RNA polymerase II promoter during meiosis, sporulation resulting in formation of a cellular spore, and regulation of DNA-template transcription (Additional File 3). The 13,500 protein clusters that were shared between Fo koae 44 and Fo 170 had no identified GO enrichment.

Using InterProScan, 16,429 and 16,487 putative protein family domains were identified in Fo koae 44 and Fo 170, respectively. Of these putative proteins, 30 were unique to Fo koae 44, and 32 were unique to Fo 170. Interestingly, a sterigmatocystin biosynthesis gene was identified as unique in Fo koae 44. Of further note is that the genome of Fo 170 also encoded some proteins involved in pathogenesis, including oxidoreductases (i.e., monooxygenase) and putrescine biosynthesis.

Using the antiSMASH database, the genes associated with secondary metabolites were similar in both genomes of Fo koae 44 and Fo 170; however, variation did exist in gene copy number (Additional File 4). Non-ribosomal peptide synthetases (NRPS), NRPS-like, and indoles had higher gene copy numbers for Fo koae 44. Fo 170 had two more copies of genes associated with the production of the Fusarium mycotoxin enniatin.

Similar results were observed from genes identified using the dbCAN2 database, except that two copies of a glycoside hydrolase, GH13_22 (characterized glucan synthase), were found in Fo koae 44, but not in Fo 170 [30]. In Fo 170, two gene copies encoding proteins GH141 (α-L-fucosidase or xylanase), one copy encoding GH30 (endo-β-1,4-xylanase or β-glucosidase), and one copy encoding GH43_12 (a glycoside hydrolase involved in hydrolysis and/or rearrangement of glycosidic bonds; arabinosidase) were identified.

Using the PHI-base, shared virulence-associated genes that were identified in both Fo koae 44 and Fo 170 include the following: transport associated genes, such as the ATP-binding cassette (Gpabc1; PHI:258); effectors, such as the Ace1 (PHI:325) and Mocdip1 (PHI:3213); Fusarium-specific transcription factors, such as the Ftf2 (PHI:5484) and the SIX gene (SGE1; PHI:3168); toxin detoxifying genes, such as the GzmetE (PHI:355) and Tom1 (PHI:438); a chitin synthase gene, CHS2 (PHI:336); and a beta-1,3-glucanosyltransferase, GAS1 (PHI:522). Nine virulence-associated genes were identified as unique to Fo koae 44. These genes encoded multiple Fusarium transcription factors, mitogen-activated protein kinases, and ATP-binding cassette (ABC) transporters. The SIX1 and SIX6 genes, were identified on the LSX of Fo koae 44, but were absent in the Fo 170 genome.

These Fo koae 44 virulence-associated genes showed sequence similarity to other F. oxysporum formae speciales (GenBank Accession numbers in Additional Files 10 and 5B). The SIX1 gene of Fo koae 44 showed highest sequence similarity with F. oxysporum f. sp. fragariae (85%) and the SIX6 gene showed highest sequence similarity to F. oxysporum f. sp. pisi (97%) and F. oxysporum f. sp. cucumerinum (97%). There were seven virulence-associated genes identified as unique to Fo 170 that consisted of signaling proteins, ATP binding cassette transporter, and one of unknown function but related to developing root rot symptoms (PEP2).

Fo koae-specific PCR primer development

To increase the likelihood of developing a pathogen-specific primer, two approaches were used to classify genomic regions and genes unique to Fo koae 44 for PCR primer development. First, unique genomic regions, that may be non-coding DNA sequences, were identified in Fo koae 44. Second, predicted transcripts unique to Fo koae 44 were identified.

To identify unique genomic regions, Fo koae 44 chromosomes were aligned to chromosomes of Fo 170, other Fusarium spp., and F. oxysporum formae speciales obtained from GenBank (Additional File 9). Analysis of these alignments identified a total of 445 DNA sequences unique to Fo koae 44. Of these 445 DNA sequences, 33 had no BLAST hits to the NCBI database, and these 33 DNA sequences were selected for design of PCR primers.

To identify unique genes, Fo koae 44 and Fo 170 were analyzed using OrthoVenn2. One thousand seven hundred sixty-two predicted proteins were identified as exclusive to Fo koae 44. The corresponding transcripts were BLASTed to the NCBI database and 137 of these transcripts had no BLAST hits. These 137 transcripts were also selected for design of PCR primers.

Taken together, each of the 33 unique genomic DNA sequences and the 137 predicted transcripts (ca. 170 genome regions) were BLASTed against the chromosomes to determine which chromosome contained the selected sequence target for potential amplification. From these sequences, 3750 primers were developed in silico. Of those, 35 primer pairs had no non-target hits in silico through NCBI primer BLAST. These primers were tested on characterized pathogenic Fo koae isolates and non-pathogenic F. oxysporum (Table 2), F. commune (FO21 and 85), and F. proliferatum (1 and 80) isolates; all but one F. commune isolate (FO21) were collected from koa. Of the 35 primer pairs, 7 (one designed to amplify the “core” chromosome 2 and 6 designed to amplify the LSX) did not amplify DNA from non-target species. The 6 primers designed to amplify the LSX did not result in amplification using DNA from Fo 170 or other characterized non-pathogenic F. oxysporum, F. commune, and F. proliferatum isolates. The primer developed on the “core” chromosome 2 (P4) resulted in amplification of DNA from two of the characterized non-pathogenic isolates (72 and 81). Of the primer pairs that did not amplify non-target Fusarium species, 2 of primer pairs P4 (“core”) and P6 (LSX) were individually subjected to BLAST against the NCBI database to further test primers in silico for specificity (Additional File 5A). The “core” chromosome primer pair P4 had BLAST hits to other Fusarium spp. P4 forward and reverse primers had hits of 20 out 20 nucleotides to three genomes of F. xyrophilum, and 15 out of 20 nucleotides to F. fujikuroi and F. graminearum genomes, indicating this genomic region is likely found in other Fusarium spp. The LSX primer pair P6 also had partial BLAST hits to other Fusarium spp. in GenBank, including F. venenatum, F. fujikuroi, and F. graminearum, and the forward and reverse primers had 19 and 11 out of 20 nucleotides hits, respectively, to F. culmorum (Additional File 5B). Therefore, further testing of these primers, P4 and P6, on a broader suite of Fusarium spp. is warranted.

Table 2 Fusarium oxysporum f. sp. koae and F. oxyporum isolates that have been screened for pathogenicity through greenhouse virulence assays at the Hawai’i Agriculture Research Center
Full size table

Field isolate collection

To test the specificity of primers P4 and P6 on a novel population of Fusarium isolates, root samples were collected from symptomatic koa trees. Root samples from 13 of the 14 sites surveyed in Hawaiʻi yielded Fusarium isolates. A total of 359 Fusarium isolates, including F. oxysporum, F. solani, F. proliferatum, F. concolor, F. lateritium, F. commune, and F. fujikuroi (based on tef1 sequence data; GenBank Accession numbers and locations in Additional File 11), were collected from 100 symptomatic trees (70.4% of surveyed symptomatic trees) across the three islands (Additional File 6). Most of the Fusarium isolates were obtained from Maunawili, Kalopa, Wood Valley, and Pa’auilo, respectively. The least represented sites were Kona, Wung Ranch, and Kaiwiki, respectively.

Genetic characterization of field-collected isolates

To genetically characterize field-collected isolates with known virulence isolates, the rpb2 was sequenced for isolates from individual trees. Sequencing of rpb2 was conducted on 100 isolates, including 16 characterized for pathogenicity (4 highly virulent, 6 moderately virulent, 1 low-virulence, and 5 non-pathogenic isolates) and 84 field-collected isolates (Additional File 7). Variation at the rpb2 locus was used to construct a TCS haplotype network that resulted in 12 haplotypes (Fig. 3).

Fig. 3
figure 3

RNA polymerase II second largest subunit (rpb2) haplotype network of 12 haplotypes identified from 100 Fusarium oxysporum and F. oxysporum f. sp. koae isolates including 84 untested, field-collected isolates and 16 characterized isolates of high, moderate, and low virulence and non-pathogenic. All highly virulent isolates clustered in the Hap_3 haplotype. Moderate and low virulence isolates clustered in the Hap_1 haplotype. The characterized non-pathogenic isolates clustered in the Hap_2, Hap_4, and Hap_6 haplotypes. The rpb2 haplotype network is overlaid with amplification from “core” chromosome primer pair P4 (designated by light green) and LSX primer pair P6 (designated in pink). Only the Hap_3 haplotype yielded amplification product with the LSX primer pair P6 from DNA of all highly virulent F. oxysporum f. sp. koae (Fo koae) and six uncharacterized isolates. Every isolate that amplified with the LSX primer pair also amplified with the “core” primer pair. “Core” chromosome primers (P4) produced amplification from three haplotypes including all highly virulent isolates in Hap_3 and two non-pathogenic isolates in Hap_4

Full size image

The haplotype network of the rpb2 revealed that the characterized highly virulent isolates (44, 77, 78, 79, 90, and 166) belonged to the same haplotype, Hap_3 (Fig. 3). Moderately virulent isolates clustered within haplotypes Hap_1 and Hap_3, while the low-virulence isolate clustered in Hap_1. Non-pathogenic isolates clustered within four haplotypes, Hap_1, Hap_2, Hap_4, and Hap_6. The developed primer pair P4, developed to amplify DNA from “core” chromosome 2, amplified DNA from 3 of the 12 rpb2 haplotypes (Hap_1, Hap_3, and Hap_4), whereas the developed primer pair P6, developed to amplify LSX DNA, only amplified DNA from one haplotype (Hap_3) (Fig. 3). P4 produced amplification of DNA from highly virulent and moderately virulent isolates; however, P4 also amplified DNA from two non-pathogenic isolates (72 and 81) in Hap_4. DNA from uncharacterized isolates in Hap_1, Hap_3, and Hap_4 was amplified using the P4 primer pairs. The P6 primer pair did not produce amplification using DNA from non-pathogenic or lowly virulent isolates. The P6 primer pair amplified DNA from all but one uncharacterized isolate in Hap_3.

Discussion

Plant pathogenic fusaria have been assessed for interspecific genetic variation to identify genomic features important for plant host infection [1, 2]. In addition, intraspecific genetic variation is a powerful tool to identify potential mechanisms for disease development by pathogenic fungi [31, 32]. This study presents a whole genome comparison between two F. oxysporum isolates collected from koa roots, one pathogenic (Fo koae 44) and one non-pathogenic (Fo 170) isolates. Through whole genome sequencing, genomic features were assessed bioinformatically for features that distinguish the pathogen from the non-pathogen. Similar to other intraspecific comparison studies, the two genomes in this study were very similar [3]. They shared 96% average nucleotide identity (ANIm), indicating that they belong to the same species [33]. The constructed whole-genome phylogeny also confirmed that these isolates were among the F. oxysporum species complex (Additional File 1). Synteny analysis identified that the putative chromosomes in both isolates were syntenic amongst the 11 shared chromosomes of the well-characterized F. oxysporum f. sp. lycopersici isolate [18]. Analysis of the syntenic chromosomes coincided with a recent study that observed higher recombination at the telomeres and amongst smaller chromosomes, while larger chromosomes were more conserved [28]. A non-syntenic “chromosome” was identified in both genomes, which is likely where most of the genes necessary to differentiate the koa pathogen are localized (Fig. 1b). Genome size has been shown to be a putative indicator of pathogenicity in filamentous fungi. Compared to their non-pathogenic counterparts, pathogens generally utilize larger genomes, which is associated with the presence of conditionally dispensable, lineage-specific accessory chromosomes [19]. These accessory chromosomes are known to be ca. 2 Mb, and they frequently house effectors necessary for disease development or evasion of host detection [4]. Despite these reports, this study found that the non-pathogen (Fo 170) had a slightly larger genome compared to the pathogen (Fo koae 44) at 50 and 48 Mb, respectively. Though Fo 170 had a larger genome, this study identified unique genes and DNA sequences, most notably the well-characterized secreted in xylem effectors (SIX genes), that were exclusive to Fo koae 44. Most of these unique genomic features of Fo koae 44 were identified amongst the LSX.

As F. oxysporum does not have an observed teleomorphic life stage, it relies on asexual means of recombination. As such, conditionally dispensable, accessory chromosomes, and horizontal gene transfer have been described as being responsible for the host-specificity of the various formae speciales (ff. spp.) [19, 34, 35]. The predominant variation between F. oxysporum ff. spp. has been found in effector profiles on these conditionally dispensable, accessory chromosomes [16]. Previous studies have suggested that repeats and transposons on the accessory chromosomes increase the occurrence of point mutations, which expands diversity in virulence genes through higher mutation and duplication rates via chromosomal rearrangements and altered gene expression through chromatin restructuring [19, 35]. The identified LSX of Fo koae 44 in this study was transposon-rich and contained important wilt-inducing effector genes, SIX1 and SIX6, providing evidence of one or more pathogen-specific, accessory chromosome(s). Interestingly, the identified LSX of Fo 170 was also transposon-rich, but it contained genes associated with root disease including the PEP2 gene. The LSX identified here likely represent a combination of multiple lineage-specific chromosomes, but due to the methodologies used in this study, the LSX likely also contains assembly contigs that did not map to the reference, but could still belong on core chromosomes. However, a portion of DNA found within the LSX likely represents one or more lineage-specific chromosome(s) for both Fo koae 44 and Fo 170. Future karyotyping studies could determine presence and quantity of accessory chromosomes. Virulence-associated genes within the LSX are likely the determining factors that distinguishes pathogenicity of Fo koae 44 on koa.

Even though 1353 and 2369 predicted proteins were identified on the LSX of Fo koae 44 and Fo 170, respectively, the function of 58 and 50% of genes for Fo koae 44 and Fo 170, respectively, could not be determined from the available databases used for this study (Additional File 8). Perhaps these genes are highly specific to Fo koae 44 and Fo 170 and are possibly important for pathogenicity but have yet to be described. Since we did not conduct transcriptomic analyses in this study, we could not validate expression of these genes. Amongst the identified genes, most virulence-associated genes, secondary metabolites, and carbohydrate active enzymes were found to be similar for Fo koae 44 and Fo 170. Several well-characterized, virulence-associated genes (e.g., Gpabc1, GzmetE, Tom1, Ace1, Mocdip1, Ftf2, and SGE1) were identified in both Fo koae 44 and Fo 170. Interestingly, these genes suggest that Fo koae 44 and Fo 170 both have the capacity for host recognition and disease development, even though Fo 170 does not cause wilt disease on koa [36].

Of the 57 and 59 unique predicted genes from the LSX of Fo koae 44 and Fo 170, respectively, the major distinctions found in Fo koae 44 were a mitogen-activated protein kinase pathway (associated with genes Fmk1, Ste11, and Ste7), which were previously identified as important for host penetration and pathogenicity, and the SIX1 and SIX6 [37]. SIX genes have been well-characterized as important host-specific pathogenicity factors in various F. oxysporum formae speciales [15, 18]. SIX1 and SIX6 genes together have been reported in F. oxysporum f. sp. lycopersici, F. oxysporum f. sp. melonis, F. oxysporum f. sp. niveum, F. oxysporum f. sp. canariensis, F. oxysporum f. sp. phaseoli and F. oxysporum f. sp. cubense [3, 16, 20, 38,39,40]. With the banana pathogen, F. oxysporum f. sp. cubense (Fo cubense), tropical race 1 and tropical race 4 can be distinguished by SIX genes [41, 42]. The same SIX1 and SIX6 genes reported here in Fo koae 44 have been previously reported in Fo cubense race 1 [5]. SIX genes are surrounded by repeat- and transposon-rich regions that increase the chances of mutation and duplication of these important effectors [19]. Our results concur with these previous observations because we observed the localization of the two SIX genes amongst the transposon-rich LSX of Fo koae 44 [29]. These SIX genes, in conjunction with their associated transcription factor (SGE1) and the mitogen-activated protein kinase pathway, may be key characteristics that contribute to pathogenicity on koa.

Although Fo 170 lacked SIX genes, a stronger putative association was observed for secondary metabolite production and transport. The identification of the trehalose, putrescine, cyclodipeptide, and monooxygenase genes in Fo 170 signified that toxin production may be more important for disease development on a potential non-koa host of Fo 170. Interestingly, another secreted peptide, PEP2 (PHI:224), was identified as exclusive to Fo 170, and it has been found to be important for development of root rot on pea. These results also suggest that while Fo 170 does not cause koa wilt disease, it may be a pathogen on another host. To differentiate and monitor the koa wilt pathogen from its non-pathogenic counterparts, pathogen-specific primers were developed.

Pathogen-specific primers have previously been shown to be effective at detecting and monitoring pathogenic strains of fungi [43,44,45,46,47], and they are powerful tools in disease management, including disease resistance-breeding programs. Similar to recent studies, developed primers were based on unique non-coding sequences and exclusive predicted genes of Fo koae 44 [43,44,45]. To reduce the risk of false positive amplification of other F. oxysporum formae speciales isolated from koa, we chose to develop primers that would amplify genes unique to Fo koae 44 that did not have identified gene ontology terms.

Due to the importance of lineage-specific DNA and the virulence-associated genes found within, the LSX was targeted for primer development to detect the presence of the lineage-specific DNA that contains putative genes necessary for development of koa wilt disease [3, 16]. We observed that primer pair P6 developed for the LSX was more reliable to distinguish pathogenic isolates than unique genes identified on the “core” chromosomes, because no non-target amplification of DNA from characterized non-pathogenic isolates was observed with the LSX primer pair (P6). Of the 12 identified rpb2 haplotypes, primer pair P4 resulted in target amplification of three haplotypes (Hap_1, Hap_3, and Hap_4), but primer pair P6 produced target amplification of only one haplotype (Hap_3). These results indicate that isolates in haplotype Hap_3 may possess some putative lineage-specific DNA necessary for development of koa wilt disease. DNA of haplotypes that only amplified with primer pairs P4, may contain a homologous gene in their “core” chromosomes that may not be required for virulence that is also shared with other Fusarium spp.

The LSX primer pair (P6) resulted in amplification of DNA from only one rpb2 haplotype which suggests that it may be too specific to amplify all highly virulent isolates within the Fo koae population; however, primer pair P6 produced amplified product for all pathogenic Fo koae isolates used in HARC’s disease resistance breeding program. The seven Fo koae isolates used in HARC’s disease resistance breeding program were collected on Hawai‘i (44 and 166), Kauai (76, 77, 78, and 79), and Oahu (90); DNA from all of these isolates was amplified with both developed primer pairs [23, 48]. For the ongoing greenhouse pathogenicity trials, these highly virulent isolates (44, 77, 78, and 79) and three moderately virulent isolates (76, 90, and 166) are combined evenly by weight and used in combination as a soil inoculum [23]. This methodology (mixing of inoculum from multiple isolates), may allow anastamoses and/or asexual genetic recombination to occur between isolates. Hypothetically, the putative lineage-specific DNA used to develop our primers could have been transferred horizontally among isolates during the inoculation trials conducted by HARC [7], which could have contributed to the observed amplification of a single haplotype with the developed P6 primer pair. Targeted studies of the Fo koae population are needed to better understand the role of horizontal gene transfer in pathogenicity.

Future studies should include greenhouse virulence assays to characterize field-collected isolates that amplified with both the core primer pair (P4) and the LSX primer pair (P6). Population genomic analyses are needed to determine genetic relationships among pathogenic Fo koae and non-pathogenic F. oxysporum isolates across the Hawaiian archipelago. Such studies can determine if a direct comparison of one pathogen and one non-pathogen is sufficient for developing a primer to detect pathogenic Fo koae, or if whole genome sequences of multiple haplotypes are needed to develop more robust, pathogen-specific primers. In addition, karyotyping could validate the presence and quantity of the putative lineage-specific chromosome(s) of Fo koae 44.

Conclusion

This study used whole genome analyses to predict virulence-associated genes that may be necessary for the development of koa wilt disease by Fo koae. This information was used to develop PCR primers for distinguishing highly virulent Fo koae isolates from non-pathogenic F. oxysporum isolates. Previously, the SIX1 and SIX6 genes have been characterized in other pathogens for their role in causing wilt diseases in other hosts, and it appears that these genes are also necessary for Fo koae to cause koa wilt disease. Genomic comparisons between Fo koae 44 and Fo 170 identified Fo 170-exclusive genes, including pathogenicity-related genes, which suggests that Fo 170 may be pathogenic to a different host without necessarily causing a wilt disease. Putative lineage-specific chromosomes were identified, but further research is needed to characterize and quantify these chromosomes. Pathogen-specific primers were developed that only amplified DNA of characterized, pathogenic isolates of Fo koae.

Because virulence assays have not yet been conducted on Fo koae haplotype isolates from the field collection, further research is needed to assess the robustness of the developed primers and determine if the predicted virulence-associated genes are required for pathogenicity. These results provide the framework for understanding the pathogen genes necessary for development of koa wilt disease and determining genetic variation of Fo koae populations across the Hawaiian Islands.

Methods

Greenhouse-characterized pathogenic and non-pathogenic isolates of Fusarium oxysporum

For this study, Fusarium oxysporum isolates that were previously screened for pathogenicity through greenhouse trials were provided by the Hawai‘i Agriculture Research Center (HARC) [48] (Table 2). These isolates were screened at HARC using methods described in Dudley et al. [48]. At 90 days post-inoculation of koa seedlings under greenhouse conditions, isolates were classified as non-pathogenic (< 10% mortality), low virulence (10–30% mortality), moderate virulence (30–70% mortality), or high virulence (70–100% mortality) (Table 2).

Twelve characterized pathogenic F. oxysporum isolates of high (44, 77, 78, and 79), moderate (8, 17, 20, 76, 90, 166, and 0540 K), and low (34) virulence, and five non-pathogenic (27, 53, 72, 81, and 170) F. oxysporum isolates were used for reference in later analyses. Isolates 44, 76, 77, 78, 79, 90, and 166 are currently used in HARC’s disease resistance screening trials. Pathogenic isolate 44 and non-pathogenic isolate 170 were selected for whole genome sequencing.

Whole genome sequencing and assembly

In preparation for DNA extraction, one isolate pathogenic to koa (Fo koae 44) and one non-pathogenic isolate (Fo 170) were first grown on ¼-strength PDA at 25 °C for 3 days. Hyphal tips were transferred to 100 ml of potato dextrose broth and shaken at 70 rpm for 7 days at room temperature. Mycelium was collected using a Buchner funnel and filtered through a Whatman® No. 1 filter paper under vacuum. Mycelium was placed in 2-ml tubes and frozen at 75 °C for 24 h prior to extraction. DNA was extracted using a cetyl trimethyl ammonium bromide (CTAB) extraction protocol adapted from Cubero et al. [49]. Tissue preparation was modified from the original protocol in that 0.1 g of frozen tissue was pulverized using a FastPrep-24Tm (M.P. Biomedicals LLC, Santa Ana, CA, USA) at 5x speed for three 20-s runs. Between runs, samples were kept frozen using liquid nitrogen. On the third run, 750 μl of CTAB extraction buffer (1% w/v CTAB; 1 M NaC1; 100 mM Tris; 20 mM EDTA; 1% w/v polyvinyl polypyrolidone) was added to each sample. Samples were incubated for 30 min at 70 °C in a heat block. One volume of chloroform:isoamyl alcohol (24:1) was added to each sample and mixed on a shaker for 20 min then centrifuged at 10,000 x g for 5 min. Upper aqueous phase (600 μl) was transferred to a new 2-ml tube. Two volumes of Precipitation buffer (two volumes; 1200 μl) was added to sample, which was mixed in a shaker for 5 min, then centrifuged at 13,000 x g for 15 min. Supernatant was removed and re-suspended in 350 μl 1.5 M NaCl. To each sample, 2 μl of 10 mg/ml RNase was added, followed by incubation at 37 °C for 30 min. One volume of chloroform:isoamyl was added to each sample and mixed on a shaker for 20 min, then centrifuged at 10,000 x g for 5 min. Upper phase was transferred to a new 1.5-ml tube and 0.6 volume of ice-cold isopropanol was added to each sample. Samples were mixed by inversion and incubated at − 20 °C for 1 h. After incubation, samples were centrifuged at 14,000 x g for 5 min at room temperature. Supernatant was removed and 1 ml 70% ethanol was added to each sample. After 1-min incubation at room temperature, samples were centrifuged at 14,000 x g for 3 min at room temperature. Pellet was dried and resuspended in 50 μl TE buffer (10 mM Tris-HCl, 1 mM EDTA). Extracted DNA was gel electrophoresed in a 2% agarose gel and quantified using a Qubit™ fluorometer (Invitrogen, Carlsbad, CA, USA). A Zymoclean™ Gel DNA Recovery Kit (Zymo Research, Irvine CA, USA) and a Zymo clean and concentrate column were used to recover and purify the DNA. Extracted DNA was sent to Macrogen (Seoul, South Korea) for TruSeq Illumina shotgun sequencing with reads of 2x151bp.

De novo genome assemblies of both Fo koae 44 and Fo 170 were constructed using SPAdes 3.11.1 [50] using default parameters. The SPAdes genome assembly was assessed for quality using QUAST [51]. Contigs with than 100x coverage were removed, and the remaining contigs were checked for similarity with Fusarium spp. sequences using the National Center for Biotechnology Information (NCBI) BLAST to screen for presence of bacterial DNA in samples.

Phylogenetic analysis

A whole-genome, maximum likelihood phylogeny was constructed to elucidate evolutionary relationships among Fo koae 44, Fo 170, other Fusarium spp., and other F. oxysporum formae speciales. REALPHY 112 [52] was used to generate whole genome alignments including the F. oxysporum f. sp. lycopersici reference genome and Magnaporthe oryzae (Both obtained from GenBank; Additional File 9). PhyML 3.1 [53] using bowtie2 2.3.4.2 [54] and the general time reversal (GTR) substitution model [55], was used to construct the whole genome phylogeny with bootstrap support (pseudoreplicates = 200) and M. oryzae as the outgroup. Whole genomes of Fusarium spp. and F. oxysporum formae speciales were obtained from the NCBI GenBank database (Additional File 9).

Genome annotation and analysis

To determine overall sequence similarity of Fo koae 44 and Fo 170 isolates, average nucleotide identity was calculated using Pyani [56]. R package ‘coRdon’ was used to determine codon usage bias [57, 58]. TransposableELMT, a wrapper script for transposable element identification, and creation of a comprehensive repeat library using RepeatModeler, RepeatClassifier, LTR_finder, ltr_harvest, and TransposonPSI (http://transposonpsi.sourceforge.net) and subsequently RepeatMasker with this library were used to identify transposable elements and repeat-rich regions for both Fo koae 44 and Fo 170 [59,60,61].

To identify unique sequences of Fo koae 44 and Fo 170, putative chromosomes were constructed using a closely related reference genome, F. oxysporum f. sp. lycopersici. Paired reads of the genomes of Fo koae 44 and Fo 170 were trimmed using BBDuk (decontamination using kmers; part of BBTools package) [62]. The trimmed reads were mapped to the reference genome using Geneious (Geneious Prime 2019.2, http://www.geneious.com/) progressiveMauve genome mapper software at medium-low sensitivity [63, 64]. Low and high coverage sites were removed using Geneious’ built-in program. Variants and SNPs were identified using FreeBayes [65] within Geneious. Variant calling format (VCF) was exported for use in comparison of SNPs and indels between Fo koae 44 and Fo 170. Synteny maps and MUMmer alignments were made using the synteny mapping and analysis program (SyMAP) [66] to create putative chromosome contigs based on the reference genome.

The de novo assembled genomes were annotated for both Fo koae 44 and Fo 170 using the MAKER 2.31.8 annotation pipeline [67] with RepeatMasker 4.0.8 [60]. A F. oxysporum-specific repeat library was constructed using RepeatModeler 1.0.11 [68] to mask interspersed repeats and low complexity DNA sequences. Three gene predictors were used in the pipeline: GeneMark-ES [69], SNAP [70], and AUGUSTUS [71]. Fusarium graminearum was used as a species model for AUGUSTUS. To identify tRNA genes, tRNAscan-SE 1.3.1 was used with default settings.

Predicted transcripts (≥150 bp) and proteins (≥50 amino acids) were analyzed using five databases for analysis of putative genes: OrthoVenn2 for putative non-orthologous proteins [72], InterProScan for protein family domains [36], antiSMASH for putative secondary metabolites [73], dbCAN2 for putative carbohydrate-active enzymes [30], and PHI-base for putative virulence-associated proteins [74]. OrthoVenn2, antiSMASH, and dbCAN2 databases’ online servers were used to analyze the predicted genes at default settings. A local BLAST was used for InterProScan and the PHI-base database. Selection of putative proteins from PHI-base was based on a ≥ 85% grade (metric of combining the query coverage, e-value, and identity values for each hit with weights of 0.5, 0.25 and 0.25, respectively). Putative virulence-associated proteins were tested for exclusivity to Fo koae 44 by aligning these sequences to sequences obtained from the NCBI database using MUSCLE in Geneious [75] (Additional File 10).

Primer development

Putative chromosomes from SyMAP were aligned using progressiveMauve [63] in Geneious. Using these alignments, unique sequences of Fo koae 44 were identified and selected. Also, predicted transcripts specific to Fo koae 44 were identified by comparing the predicted transcriptomes of Fo koae 44 and Fo 170 from the Maker annotation identified from Orthovenn2. Fo koae 44 predicted transcripts were further identified as unique by using a local BLAST against Fo 170 and the NCBI database. Sequences with no sequence similarity to sequences from Fo 170, other Fusarium spp., and F. oxysporum formae speciales were used to develop primer pairs. Primer3 [76] was used to develop PCR primers. Fo koae 44 unique sequences and predicted transcripts were used to generate ca. ten primer sets (forward and reverse primers) for each sequence and transcript.

Primers were tested for specificity using template DNA from four characterized highly virulent Fo koae isolates (isolates used in HARC greenhouse screening trials: 44, 77, 78, and 79), seven characterized moderately virulent Fo koae isolates (HARC isolates 8, 17, 20, 76, 90, 166, and 0540 K), two characterized non-pathogenic Fo isolates (HARC isolates 27 and 170), two F. commune isolates (HARC isolate 85 and Stewart collection FO21), and one F. proliferatum isolate (HARC isolate 1). Each 25-ul reaction contained 10 ng template DNA or sterile, molecular grade water for the negative control. Primers were tested using a Mastercycler ProS thermocycler (Eppendorf, Hamburg, Germany) at a program of 94 °C for 2 min, 30 cycles of 94 °C for 40 s, 59 °C for 40 s, and 72 °C for 30 s, and 72 °C for 5 min. PCR products for each primer set and template DNA were electrophoresed using a standard 1.5% agarose gel at 60 V for 60 min, and visualized using GelRed® (Biotium, Freemont, CA, USA).

Field collection of Fusarium spp.

To assemble an expanded population of F. oxysporum isolates for Fo koae specific-primer tests, 14 collection sites were selected based on a previous survey conducted by HARC from 2004 to 2007 across four of the main Hawaiian Islands, including Kauai, Oahu, Maui, and Hawai‘i [77]. Sites for new isolate collections were selected based sites on where previously characterized pathogen isolates of moderate to high virulence were collected. In addition, selected collection sites were geographically distant from each other to increase the likelihood of collecting a more genetically diverse pathogen population. On Kauai, the three collection sites included Kokee, Kapa’a, and Hanalei (Princeville). On Oahu, three collection sites included Maunawili, Poamoho, and Kahana. On Hawai‘i, eight collection sites included Wood Valley, Pu‘u wa‘a wa‘a, Kona, Wung Ranch, Kaiwiki, Kalopa, Paauilo, and Volcano National Park. Root samples were collected from symptomatic trees at each site. Symptomatic segments of 4- to 5-mm thick roots exhibiting discoloration and black streaking, were targeted for isolation of Fusarium spp. Five fine roots and/or pencil-thick roots were plated on selective Komada media [78] for isolation of Fusarium spp. Plates were incubated at 25 °C for 3 days then checked for mycelial growth emanating from the root samples. Hyphal tips were taken from the radiating mycelium and transferred to ¼-strength PDA and incubated at 25 °C for another 3 days. Colony morphology was used to determine if the isolate was Fusarium. To further characterize these isolates to species, isolates were sequenced at the translation elongation factor 1 alpha (tef1) region using a Mastercycler ProS thermocycler and a PCR cycle program of 94 °C for 2 min, 40 cycles of 94 °C for 40 s, 58 °C for 40 s, and 72 °C for 30 s, and 72 °C for 5 min [79]. Products were run on a 1.5% agarose gel to visualize amplified PCR product using GelRed®.

Genetic characterization of field-collected isolates

To determine the genetic relationships among the field-collected isolates and the previously characterized pathogenic and non-pathogenic isolates, F. oxysporum isolates, based of tef1 data. Isolates were selected from separate, individual trees. The selected isolates were then sequenced at the RNA polymerase II second largest subunit (rpb2) [80] using previous methods above for tef1. Rpb2 sequences of both characterized and field-collected isolates were aligned using MUSCLE alignment in Geneious [75]. A statistical parsimony (TCS) haplotype network was constructed based on this rpb2 sequence data using PopART with a 95% genetic cut-off [81]. These isolates were tested with the developed Fo koae-specific primers to identify haplotypes that may contain the putative lineage-specific DNA identified from Fo koae 44.