Genome sequencing of oomycete isolates from Chile supports the New Zealand origin of Phytophthora kernoviae and makes available the first Nothophytophthora sp. genome

Summary Genome sequences were generated for six oomycete isolates collected from forests in Valdivia, Chile. Three of the isolates were identified morphologically as Phytophthora kernoviae, whereas two were similar to other clade 10 Phytophthora species. One isolate was tentatively identified as Nothophytophthora valdiviana based on nucleotide sequence similarity in the cytochrome oxidase 1 gene. This is the first genome sequence for this recently described genus. The genome assembly was more fragmented and contained many duplicated genes when compared with the other Phytophthora sequences. Comparative analyses were performed with genomic sequences of the P. kernoviae isolates from the UK and New Zealand. Although the potential New Zealand origin of P. kernoviae has been suggested, new isolations from Chile had cast doubt on this hypothesis. We present evidence supporting P. kernoviae as having originated in New Zealand. However, investigation of the diversity of oomycete species in Chile has been limited and warrants further exploration. We demonstrate the expediency of genomic analyses in determining phylogenetic relationships between isolates within new and often scantly represented taxonomic groups, such as Phytophthora clade 10 and Nothophytophthora. Data are available on GenBank via BioProject accession number PRJNA352331.


SUMMARY
Genome sequences were generated for six oomycete isolates collected from forests in Valdivia, Chile. Three of the isolates were identified morphologically as Phytophthora kernoviae, whereas two were similar to other clade 10 Phytophthora species. One isolate was tentatively identified as Nothophytophthora valdiviana based on nucleotide sequence similarity in the cytochrome oxidase 1 gene. This is the first genome sequence for this recently described genus. The genome assembly was more fragmented and contained many duplicated genes when compared with the other Phytophthora sequences. Comparative analyses were performed with genomic sequences of the P. kernoviae isolates from the UK and New Zealand. Although the potential New Zealand origin of P. kernoviae has been suggested, new isolations from Chile had cast doubt on this hypothesis. We present evidence supporting P. kernoviae as having originated in New Zealand. However, investigation of the diversity of oomycete species in Chile has been limited and warrants further exploration. We demonstrate the expediency of genomic analyses in determining phylogenetic relationships between isolates within new and often scantly represented taxonomic groups, such as Phytophthora clade 10 and Nothophytophthora. Data are available on GenBank via BioProject accession number PRJNA352331.
Keywords: Drimys winteri, Eucalyptus nitens, forest disease, hybrid, Nothophytophthora, oomycete, Phytophthora Pests and diseases, together with climate change, present the biggest threats to forest health (Ramsfield et al., 2016;Trumbore et al., 2015). The abundance and diversity of Phytophthora species known to be negatively impacting forest trees are increasing (Hansen, 2015;Scott and Williams, 2014;Studholme et al., 2016). The knowledge of Phytophthora species diversity within forests is limited (Hansen et al., 2012), although momentum is building to characterize these populations phylogenetically and to better understand their biology and impacts on forest health (Burgess et al., 2017;Scarlett et al., 2015). Tree diseases are likely to be ongoing and permanent features of the Anthropocene forest landscape as a result of human activities (Potter and Urquhart, 2017) with economic, landscape and societal impacts (Drake and Jones, 2017). For these reasons, it is essential that we understand the introduction pathways and how new pathogen populations emerge, and establish ways to mitigate the impacts of forest diseases.
Phytophthora kernoviae was first described in 2003, causing stem cankers on forest trees and foliar lesions on ornamentals in the UK (Brasier et al., 2005). Subsequent studies discovered P. kernoviae in New Zealand, where it may have been present since at least the 1950s (Ramsfield et al., 2007). In New Zealand, P. kernoviae is not considered to be a major pathogen of native flora, but has greater pathogenicity on exotic plant hosts (Gardner et al., 2015). Phytophthora kernoviae has also been co-isolated from needles exhibiting red needle cast in New Zealand's Pinus radiata plantations, but not as frequently as Phytophthora pluvialis (Dick et al., 2014), and not in association with substantial disease or mortality (Scott and Williams, 2014). The absence of symptoms on native host plants, and the long history of its presence, previously led to speculation that New Zealand may be the centre of origin for P. kernoviae. Phytophthora kernoviae has been isolated from Drimys winteri (winter's bark or canelo) in native forests in Chile (Sanfuentes et al., 2016). Unlike the situation in New Zealand, P. kernoviae has not been detected in P. radiata plantations in Chile, in places in which severe attacks of P. pinifolia normally occur (E. Sanfuentes, personal information). Previously known only in Europe (UK) and New Zealand, the discovery of P. kernoviae on a third continent, and the common ancient Gondwanaland flora shared by New Zealand and Chile, raise questions as to its true place of origin. This might be resolved by the determination of phylogenetic relationships among isolates from each geographical location.
Genomes have been sequenced for isolates of P. kernoviae from New Zealand (Studholme et al., 2016) and the UK (Feau et al., 2016;Sambles et al., 2015). Here, we report the first genome sequences for isolates of P. kernoviae from Chile (Table 1), the only other geographical location in which this species has been reported so far. Furthermore, we present the genome sequences of two unidentified isolates that belong to the Phytophthora internal transcribed spacer (ITS) clade 10 (Cooke et al., 2000), and report the first genome sequence for a member of Nothophytophthora, a newly described sister genus to Phytophthora (Jung et al., 2017). This Nothophytophthora isolate was collected from infected Eucalyptus nitens foliage, indicating the pathogenic potential for this genus, revealing a hitherto unrecognized potential threat to forest trees. The genetic diversity observed in these genome sequences demonstrates that the biodiversity of oomycetes in Chilean forests warrants further investigation. With increased efforts to understand the diversity and distribution of potential pathogens, there is a need to quickly establish the identity, diversity and distribution of newly described isolates. The main objective of this study was to determine the relationship of newly discovered Chilean P. kernoviae to isolates from Europe and New Zealand. A second objective was to characterize other Phytophthora-like isolates from Chilean forests and their taxonomic and phylogenetic positions within the oomycetes to obtain a better understanding of the diversity and biosecurity implications of the species.
Phytophthora isolates were obtained from soil, leaf litter and infected E. nitens foliage samples collected from three locations in Chile (Table 1) using standard baiting and isolation techniques (Erwin and Ribero, 1996). Preliminary identification of each isolate was performed using DNA sequences of ITS-1 and ITS-4 (White et al., 1990as described in Jung et al.. (2017. For genome sequencing, DNA was extracted from mycelium using a GeneJET DNA purification kit (Thermo Fisher), and its quality was assessed using a Qubit fluorometer (Thermo Fisher, Waltham, MA, USA.) and agarose gel electrophoresis. Genomic DNA from each isolate was sequenced using an Illumina HiSeq 2500 (San Diego, CA, USA) to generate 295-bp paired-end reads. A de novo assembly and scaffolding were performed using SPAdes v3.8.1 (Bankevich et al., 2012) after first removing poor-quality data and adaptor sequences using TrimGalore (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Genomes were annotated with predicted genes identified using the MAKER pipeline v2.31. 10 (Cantarel et al., 2008), incorporating ab initio gene prediction based on AUGUSTUS v3.1.0 trained on P. kernoviae genes; full details are provided via the MAKER configuration files in Text S1 (see Supporting Information). Annotated genome assemblies and raw sequence data are available via GenBank ( Table 1). The assembly statistics for the genome sequences presented are comparable with those published previously (Feau et al., 2016;Sambles et al., 2015;Studholme et al., 2016), with the exception of Chile 5. This genome assembly had a larger number of scaffolds and appeared to be much larger in total length ( Table 1). The completeness of the genome assemblies was assessed using BUSCO v2 (Benchmarking Universal Single-Copy Orthologs) (Simao et al., 2015) to check the presence of conserved single-copy orthologous genes commonly conserved across the Alveolata-Stramenopile lineage. This analysis was performed on both the genome sequences and predicted proteomes (Table 2). Completeness for genes from Chile 1, Chile 2, Chile 4, Chile 6 and Chile 7 (95.7%-97.9%) was similar to that of recently published P. kernoviae genomes (97.4%-97.9%) (Feau et al., 2016;Sambles et al., 2015;Studholme et al., 2016) (Table 2). The Chile 5 assembly was less complete (87.6%) and had a considerably higher number of duplicated genes (164) compared with the other genomes, which had less than four duplicated genes. In addition, the Chile 5 assembly was missing more genes (25) than the other genomes (Table 2). This pattern was also reflected in the predicted proteomes and may indicate that Chile 5 is a hybrid. Similar observations have been made previously with BUSCO analysis of predicted genes by Feau et al. (2016) for the genome sequence of P. alni sp. alni, a well-characterized hybrid Phytophthora species. Jung et al. (2017) noted the presence of heterozygous nucleotides within several gene regions in their Nothophytophthora isolates, and suggested the need for further investigation of hybridization or ploidy variation.
To initially identify the species of the isolates, we extracted the cytochrome oxidase I (COI) gene sequences from the genome assemblies and performed BLASTN (Altschul et al., 1990) similarity searches against the National Center for Biotechnology Information non-redundant (NCBI NR) database. The COI sequences from Chile 1, Chile 2 and Chile 4 were 99% identical to those of other P. kernoviae sequences. Sequences from Chile 6 and Chile 7 also showed close matches to P. kernoviae, but at 97% identity (data not shown). Interestingly, the COI sequence *BUSCO coverage when tested with Alveolata-Stramenopile gene set (n = 234 genes) (Simao et al., 2015) † Reference for dataset description and BUSCO analysis with annotated genes. BUSCO analysis with annotated proteins performed in this study. ‡ -, not determined. Genomes from strains 00629/1 and 00844/4 are almost identical to that of 00238/432 (Sambles et al., 2015), and strain CBS 122049 is also very similar to the other UK strains; hence, 00238/432 was used solely to represent the UK strains.
To more robustly assess the relationships between the genomes of the sequenced isolates, the Reference Sequence Alignment-based Phylogeny Builder (REALPHY) tool was used to reconstruct the phylogenies from raw genome sequencing reads (Bertels et al., 2014). Sequencing reads from the six Chilean genomes were aligned with those of the other available P. kernoviae isolates from the UK and New Zealand (Sambles et al., 2015;Studholme et al., 2016). Chile 1, Chile 2 and Chile 4 clustered adjacent to the P. kernoviae isolates from New Zealand and the UK (Fig. 1), consistent with the phylogenies constructed using COI (Text S2; Fig. S1, see Supporting Information), indicating that these isolates were P. kernoviae. Chile 6 and Chile 7 appeared divergent to the other Chilean isolates (Fig. 1), clustering separately from P. kernoviae and other clade 10 Phytophthora species. Chile 5 clustered separately to all Phytophthora genomes in each constructed phylogeny, but clustered with Nothophytophthora in the COI alignment (Fig. S1).
To complement the REALPHY-based phylogenetic analysis, isolates were clustered according to the sequence similarity of their genome assemblies by pairwise genome alignments using the dnadiff tool in the software package MUMmer v3.23 (Kurtz et al., 2004). Average nucleotide identities (ANIs) of one-to-one alignments were obtained from dnadiff, and a heatmap of the ANI similarity matrix was generated using the BIONJ clustering method (Gascuel, 1997). This analysis supported the REALPHY results, with nucleotide sequence identities among all pairs of New Zealand and UK P. kernoviae isolates and Chilean isolates 1, 2 and 4 being greater than 99%, whereas Chile 6 and Chile 7 showed 97% identity with P. kernoviae. Furthermore, Chile 5 shared only 84% identity with the P. kernoviae isolates (Fig. 2), suggesting more distant relationships between these groups. Further support for the genetic diversity of New Zealand isolates was observed using the detected single nucleotide polymorphisms (SNPs) concatenated and aligned for a splits tree analysis (Text S2; Fig. S2, see Supporting Information). Genetic variation between the genomes of the New Zealand isolates was found to be greater than that for the UK or Chilean P. kernoviae isolates. Although this analysis would benefit from a greater number of genomes for comparison, it provides early insights suggesting that New Zealand may have greater genetic diversity and be the origin of this species.
The interspersed repeats within the genomes of P. kernoviae isolates were determined by RepeatMasker v4.0.7 (Smit et al., 2015) using a combined database of DFam_Consensus v2017-01-27 (Wheeler et al., 2013) and RepBase Phytophthora library v2017-01-27 (Jurka et al., 2005). This revealed that the content of repetitive elements for the New Zealand strains was approximately 8.6% each, whereas the Chilean and UK isolates ranged from 10.7% to 16.48% and 5.98% to 7.48%, respectively (Table S1, see Supporting Information). Overall, these results are lower than those for previously reported genomes of Phytophthora species, such as P. cactorum (46.7%) (Yang et al., 2018), P. ramorum (28%), P. sojae (38%) and P. infestans (74%) (Haas et al., 2009), but may simply reflect differences in sequencing and assembly methods, with redundant sequences being collapsed to single contigs. Of all the genomes analysed in this study, Chile 5 showed the greatest overall repetitive sequence content. Genes encoding candidate RxLR and crinkler (CRN) effectors were predicted from the P. kernoviae genomes by searching for the RxLR-EER and LFLAK-HVLV motifs from amino acid sequences (translated from all of the open reading frames in a genome) using regular expression and homology search mechanisms based on hidden Markov models (HMMs) (Eddy, 1998;Tabima and Grunwald, 2018). Elicitins were predicted on the annotated gene models based on HMM models built from lists of known elicitins on UniProt (The UniProt Consortium, 2015) and established models of primary elicitin structures (Jiang et al., 2006). The candidate effector proteins were passed through the SignalP v3.0 (Dyrløv Bendtsen et al., 2004) program, and those with a predicted signal peptide were considered as candidate effectors. The numbers of predicted RxLR, CRN and elicitin effectors were similar for all P. kernoviae isolates, with the Chilean genomes showing slightly higher numbers of RxLRs compared with the other P. kernoviae genomes. By stark contrast, the genome from isolate Chile 5 revealed only five RxLR proteins compared with all other genomes which contained 116-159 RxLR proteins, and a much higher number of elicitins (88) compared with the others (<21) (Fig. S3, see Supporting Information). These results are consistent with the taxonomic and phylogenetic distinctness of Chile 5 compared with all other Chilean isolates, and highlights the diversity of virulence effectors and genetic diversity to be found among Chilean Phytophthora populations.
The extent of heterozygosity was estimated in each genome by aligning genomic sequence reads against the NZFS 2646 reference genome assembly. This approach assumes that, in shotgun sequencing of a diploid genome, sequence reads are drawn randomly from each chromosome, giving an estimate of heterozygosity across the genome (Turner et al., 2017). Figure 3 shows the frequency of distributions for the most abundant and second most abundant base at each position in the genome. A highly homozygous genome would be expected to display a single peak close to 100% abundance for the most common base and a single peak close to zero for the second most abundant, with heterozygous sites contributing to a second peak close to 50% abundance representing an overlap in the most common and second most abundant base. The resulting plots of the estimated allele frequency show a pronounced peak close to 50% for the New Zealand isolate NZFS 3630 and, to a lesser extent, for NZFS 2646 and the UK isolates (Fig. 3). Given that P. kernoviae is homothallic (Brasier et al., 2005), low levels of heterozygosity amongst most isolates could be explained by multiple generations of inbreeding (Goodwin, 1997), which would be consistent with a single introduction of a founder population with narrow genetic diversity. In contrast, a relatively high degree of heterozygosity (as observed for NZFS 3630) would be expected in individuals sampled from a genetically diverse and frequently outcrossing population, as would be expected at the centre of origin. Another possibility is that isolate NZFS 3630 is drawn Fig. 2 Average nucleotide identity matrix of Phytophthora sequence identities based on alignments of whole assembled genomes. The colour key represents the pairwise percentage similarity of alignment results from the dnadiff tool in MUMmer v3.23 (Kurtz et al., 2004). [Colour figure can be viewed at wileyonlinelibrary.com] from a population of individuals which reproduce exclusively by non-sexual vegetative means. Although heterothallism has not yet been demonstrated in P. kernoviae, it is a possibility and cannot be excluded.
Gondwanan origins have been investigated for several plant pathogens, including Phytophthora cinnamomi (Arentz, 2017). The potential Gondwanan ancestry of the isolates in this study remains plausible and is in agreement with the common ancestral plant diversity of both Chile and New Zealand (McCarthy et al., 2007;Segovia et al., 2015). Our data are consistent with previous studies proposing that P. kernoviae in the UK is of New Zealand origin; Fig. 1 shows the UK isolates forming a single narrow lineage within the radiation of diversity found in New Zealand. Chilean isolates 1, 2 and 4 represent an outgroup, not originating within the breadth of diversity found in New Zealand. Such a pattern is consistent with Chilean and New Zealand populations both being derived from an ancient pan-Gondwanan population. Within New Zealand and Chile, the subtle and intermittent expression of disease symptoms on native plant hosts suggests that P. kernoviae populations have co-evolved with plant hosts in both their respective ecosystems, reflecting the similarities in plant diversity and climate between the two countries. However, genotyping of further isolates sampling the genetic diversity in New Zealand and South America is necessary to test this hypothesis. Molecular clock models may infer the timing of the last common ancestor of New Zealand and Chilean populations. Previous attempts to date events in oomycete evolution have estimated the origin of the genus Phytophthora at less than 50 million years ago (Matari and Blair, 2014), implying that speciation in Fig. 3 Frequency distributions of relative abundances of major alleles over all genomic sites in sequenced Phytophthora kernoviae genomes. For each isolate, sequence reads were aligned against the appropriate genome assembly using BWA-mem (Li and Dublin, 2009;Li, 2013). The histogram shows the frequency densities for the relative abundance of the most common nucleotide in blue and the second most abundant nucleotide in red. The histograms have been cropped at 0.1 and 0.9 for clarity. [Colour figure can be viewed at wileyonlinelibrary.com] Phytophthora took place after the breakup of Gondwana about 180 million years ago. For greater confidence, molecular clock studies of the Phytophthora genus and Peronosporales would benefit from analyses across a much greater number of isolates, especially with the inclusion of basal clade 10 Phytophthora and Halophytophthora isolates.
Current phylogenies place P. kernoviae in clade 10 (Cooke et al., 2000) of the genus, together with four other species: Phytophthora boehmeriae K. Sawada (Erwin and Robeiro, 1996), Phytophthora gallica T. Jung & J. Nechwatal (Jung and Nechwatal, 2008), Phytophthora morindae Z.G. Abad & S.C. Nelson (Nelson and Abad, 2017) and Phytophthora intercalaris (Yang et al., 2015). Phytophthora gallica has been recovered from the rhizosphere of declining oak in France and Germany, and identified as a moderately aggressive pathogen, but of lower impact than other pathogens found within the same system (Jung and Nechwatal, 2008). In contrast, P. boehmeriae has long been associated with leaf, boll and root rot of a broad range of host species dating back to 1927 (Erwin and Robeiro, 1996). The newly described species, P. morindae, causes foliar and fruit rot disease of Indian Mulberry (Morinda citrifolia L. var. citrifolia) (Nelson and Abad, 2017), whereas P. intercalaris has frequently been isolated in stream baiting surveys of eastern USA, but has yet to be associated with disease symptoms on any plant host (Yang et al., 2015). The increasing numbers of pathogens being assigned to Phytophthora clade 10, together with the known pathology of most of these species, demonstrate the potential for pathogens from this clade to cause significant disease on a range of hosts in suitable climatic conditions. The phylogenetic diversity amongst the few isolates presented here suggests that a considerable level of taxonomic diversity within clade 10 remains undescribed with implications for the emergence of new pathogens. Chilean isolates 5, 6 and 7 are clearly distinct, with Chile 5 probably being Nothophytophthora valdiviana, and Chile 6 and Chile 7 representing another separate lineage and possibly distinct species within the clade 10 Phytophthora species that may be unique to South America. The Phytophthora isolates studied here were all collected within a single region of Chile; a broader analysis of isolates from regions of a comparable climatic range to the temperate areas of New Zealand in which P. kernoviae has been isolated would be of considerable benefit.
The genome sequences presented here provide a resource for the development of genetic markers (e.g. SNPs and short repeats) that could be used for the assessment of population diversity (Brar et al., 2018). This study also highlights the power of whole-genome sequencing as a tool for the identification of Phytophthora species, including potential hybrids and yet-to-be described species.

AC K N OW L E D G E M E N T S
DJS was supported by a grant from the Gatsby Charitable Foundation. RLM, PP and NMW were funded by MBIE (CO4X1305), the Forest Growers Levy Trust (administered by the New Zealand Forest Owners Association) and the Radiata Pine Breeding Company under the 'Healthy trees, Healthy future' research programme at Scion (NZFRI, Ltd.). The authors would like to acknowledge Dr Karen Moore and the Exeter Sequencing Service for technical assistance with DNA genome sequencing.

SU PP O R T I N G I N F O R M AT I O N
Additional supporting information may be found in the online version of this article at the publisher's web site:

Fig. S1
Relationships of Chilean Phytophthora isolates based on cytochrome oxidase I DNA sequences. Phytophthora isolates from Chile are indicated with an asterisk. IMI393172 is the P. kernoviae Holotype (CBS website). GenBank accession numbers are given for sequences from previous studies. The DNA sequences were aligned using the MAFFTT plugin in Geneious (v10.2.2) and the phylogenetic tree was constructed using the RAxML plugin and edited in FigTree V1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). Fig. S2 The relationship between eleven strains of P. kernoviae (from UK; n=4, NZ; n=2, and Chile; n=5) and closely related isolates (Chile 6 and 7 from Chile). SNPs detected from across the entire genomes were concatenated, aligned and a tree constructed using SplitsTree (Huson & Bryant, 2006). A; full Splitstree depicting relationships of all genomes, B; inset from A showing detailed relationships of P. kernoviae isolates. Chile 5 was not included in this analysis. Fig. S3 Numbers of predicted RxLR, CRN and elicitin genes from Phytophthora genomes. Elicitin gene numbers were not predicted from 0069/1, 00844/4 or CBS 122049. Genomes from strains 0069/1 and 00844/4 show a very high level of identity to that of 00238/432 (Studholme et al. unpublished), and strain CBS 122049 is also very similar to the other UK strains, hence 00238/432 was used solely to represent the UK strains.