Host Adaptation and Speciation through Hybridization and Polyploidy in Phytophthora

It is becoming increasingly evident that interspecific hybridization is a common event in phytophthora evolution. Yet, the fundamental processes underlying interspecific hybridization and the consequences for its ecological fitness and distribution are not well understood. We studied hybridization events in phytophthora clade 8b. This is a cold-tolerant group of plant pathogenic oomycetes in which six host-specific species have been described that mostly attack winter-grown vegetables. Hybrid characterization was done by sequencing and cloning of two nuclear (ITS and Ypt1) and two mitochondrial loci (Cox1 and Nadh1) combined with DNA content estimation using flow cytometry. Three different mtDNA haplotypes were recovered among the presumed hybrid isolates, dividing the hybrids into three types, with different parental species involved. In the nuclear genes, additivity, i.e. the presence of two alleles coming from different parents, was detected. Hybrid isolates showed large variations in DNA content, which was positively correlated with the additivity in nuclear loci, indicating allopolyploid hybridization followed by a process of diploidization. Moreover, indications of homeologous recombination were found in the hybrids by cloning ITS products. The hybrid isolates have been isolated from a range of hosts that have not been reported previously for clade 8b species, indicating that they have novel pathogenic potential. Next to this, DNA content measurements of the non-hybrid clade 8b species suggest that polyploidy is a common feature of this clade. We hypothesize that interspecific hybridization and polyploidy are two linked phenomena in phytophthora, and that these processes might play an important and ongoing role in the evolution of this genus.


Introduction
Phytophthora is a genus of plant pathogenic filamentous oomycetes containing more than one hundred species. Virtually all of them are plant pathogens causing many important plant diseases worldwide, such as potato late blight, sudden oak death and forest dieback caused by Phytophthora infestans, Phytophthora ramorum and Phytophthora cinnamomi, respectively.
Morphologically, oomycetes are very similar to the filamentous fungi (kingdom Fungi). Therefore, they used to be classified as Fungi. Only during the last few decades, DNA and whole genome sequencing revealed that they have a completely different evolutionary origin but have adopted similar morphology and infection strategies through convergent evolution and horizontal gene transfer. Oomycetes differ from Fungi in some important morphological and biochemical aspects and in the fact that they are diploid in their vegetative lifestyle, whereas most Fungi are monoploid [1].
Phytophthora clade 8b contains a group of pathogens specifically adapted to cause disease at low temperatures in a range of important agricultural crops, mostly winter grown vegetables. A previous genetic diversity study of Phytophthora clade 8b isolates from around the world resulted in the official description of three new clade 8b species. This was based on differences in the nuclear rDNA internal transcribed spacer (ITS) and mtDNA cytochrome oxidase I (Cox1) barcoding sequences [2], morphological and physiological differences, and host preference [3]. The clade now consists of six hostspecific species, namely P. porri, P. primulae and P. brassicae and the newly described P. dauci, P. cichorii and P. lactucae, respectively causing disease in leek, primrose, cabbages, carrot, chicory and lettuce. Two additional taxa were described, which have been tentatively named P. taxon castitis (on carrot and strawberry) and P. taxon parsley (on parsley), as insufficient data could be collected for an official species description.
During this previous study [3], we detected intra-isolate sequence polymorphisms in the ITS region in 12 isolates, which were mainly isolated from Allium species. These sequence polymorphisms point to additivity, which is a distinctive feature of interspecific hybridization. Therefore, we decided to study the possible hybridity of these isolates using different techniques, which laid the foundation of this work.
Natural interspecific hybridization has already been reported several times in the genus Phytophthora [4][5][6][7][8][9][10][11][12]. Next to this, synthetic hybrids have repeatedly been created in the lab [13][14][15]. In Phytophthora, the fundamental processes underlying interspecific hybridization and the consequences for its ecological fitness and distribution are not well understood. In plants, however, it is known that hybridization and polyploidy are important evolutionary processes contributing to adaptation and speciation and form the basis of the complete (angiosperm) biodiversity known today [16].
One of the main pathways of (angiosperm) speciation is believed to involve a polyploid phase, mostly occurring as a consequence of hybridization. In a hybridization event, (slightly) diverged chromosome sets are brought together in one nucleus. This causes unstable meiosis and in most cases abortion of the zygote. Fertility can be restored by a whole genome duplication event, creating again compatible chromosome sets. The resulting allopolyploid will then harbor the complete genomes of both parents. Although allopolyploids can be stable over long periods of time, it is well documented that most (allo)polyploids evolve back to a diploid state through the process of diploidization [17]. Diploidization can be defined as the transitioning of a polyploid organism back to the more stable diploid state. This can be accompanied by chromosome number reduction, although this is not necessarily the case. Processes that contribute to diploidization include subfunctionalization and/or neofunctionalization causing differentiation between redundant gene copies; large scale chromosomal rearrangements such as translocations; chromosome loss, etc. In allopolyploids, an important process that can create new, advantageous gene combinations and phenotypes is homeologous recombination, i.e. recombination between chromosomes of different parental origin [18].
Polyploidy has already been shown in several Phytophthora species [19][20][21]. In P. infestans, Sansome [20] showed that British isolates contained approximately twice the number of chromosomes compared to isolates from Mexico, the species' center of origin. She hypothesized that P. infestans might exist in the tetraploid condition in temperate regions, and that the higher ploidy levels might enable the pathogen to adapt to cooler environments. This initiated a DNA content screening of P. infestans populations in many countries, using cytophotometric methods. Indeed, isolates from Mexico were found to contain much lower DNA contents compared to isolates from other regions [22], supporting Sansome's hypothesis.
With the advent of the genomic era around the year 2000, research efforts aimed at understanding polyploidy in Phytophthora diminished. However, in 2010, a new study with recent P. infestans field isolates analyzed using flow cytometry showed large DNA content variation and heterokaryosis [23]. Moreover, by analysis of the genomes of P. infestans, P. ramorum and P. sojae using bio-informatics, remnants of an ancient polyploidization event were detected. Most likely, a common ancestor of these species has undergone a whole genome duplication that might have played a role in the evolution and pathogenic success of Phytophthora pathogens [24].
In this paper, we describe three different types of interspecific hybrids in Phytophthora clade 8b, as well as the occurrence of polyploidy as a common feature of the clade. We discuss a potential link between polyploidy and past hybridization events and the role that both events could play in host adaptation and speciation of Phytophthora pathogens. The implications of these phenomena for Phytophthora research are discussed.

Isolate collection and maintenance
All isolates used in this study are listed in Table 1. The isolates were freshly isolated from diseased plants or obtained from different culture collections around the world. Thirty-one of these isolates have been used previously in a genetic diversity study of Phytophthora clade 8b [3]. The isolates were maintained routinely on V8 agar [3] or on Corn Meal Agar (Beckton Dickinson). For long term storage, isolates were kept on V8 plugs at -80°C in 10% glycerol.
The amplifications were done in a Flexcycler PCR Thermal Cycler (Analytikjena). For ITS and Ypt1 the following program was used: initial denaturation for 10 min at 94°C; 35 cycles of denaturation for 1 min at 94°C; annealing for 1 min at 60°C; extension for 1 min at 72°C; final extension for 10 min at 72°C. For the mtDNA genes, another program was used: initial denaturation for 10 min at 94°C; 40 cycles of denaturation for 1 min at 94°C; annealing for 30 sec at 52°C; extension for 1 min at 72°C; final extension for 10 min at 72°C.
To reduce the impact of PCR mediated recombination [25] in the ITS region as was detected in our study, an improved PCR protocol was designed following the instructions suggested by Lahr and Katz [26]. More specifically, a new forward primer (ITSPA) was designed to reduce the size of the template to around 350 bp, now containing only the last 5 polymorphic base pairs between the P. porri and P. primulae/P. taxon parsley ITS sequences. Moreover, the amount of template was reduced to the absolute minimum needed for sufficient amplification and a proofreading polymerase (Phusion, NEB) was used. The reaction was performed in a 40 µL mix containing 8 µL Phusion HF buffer (5x, NEB), 0.8 µL dNTPs  All PCR samples were held at 4°C before analysis by gel electrophoresis on 1% TAE gels containing Ethidium Bromide and evaluation under UV fluorescence. Before sequencing, PCR products were purified using the QiaQuick PCR Purification Kit (Qiagen). Both strands were sequenced by LGC Genomics (Berlin), using the same primers as for PCR.
Cloning of ITS sequences was done using the CloneJET™ PCR Cloning Kit (Fermentas). Ten to twenty clones per isolate were sequenced for analysis. All newly generated sequence data have been submitted in GenBank (KF882609 -KF882735).

Phylogenetic analysis
Alignments for the nuclear Ypt1 and for the two mitochondrial (Cox1 and Nadh1; appended) sequences, were made using MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/), and manually edited afterwards using BioEdit. The two mtDNA genes represent about 4% of the mitochondrial genome, taking P. ramorum (clade 8c) as a reference for the mitochondrial genome size [27]. Phylogenetic analysis was done with the Maximum Likelihood algorithm using MEGA 5.2 [28]. Model testing was done using the software implemented in MEGA 5.2. For the mtDNA (Cox1 and Nadh1) alignment, the HKY + G + I substitution model was chosen and for Ypt1, the K2 + I model was used. Bootstrapping was done with 1000 replicates. All alignments and phylogenetic trees are publicly available in Treebase (study number S14964; study url: http://purl.org/ phylo/treebase/phylows/study/TB2:S14964) DNA content analysis by flow cytometry. DNA contents were measured by flow cytometry for all isolates listed in Table  1, except CBS 112969. Phytophthora isolates were taken from storage at -80°C and grown in test tubes containing 5 mL clarified V8 broth for 6-10 days. The mycelium was then harvested and washed three times with sterile water. Raphanus sativus cv. Saxa was chosen as the internal DNA reference standard (with a genome size of 2C = 1.11 pg [29]). Nuclei extraction was done using the Cystain PI absolute P kit (Partec, Münster, Germany). For each sample approximately 0.5 cm 2 of young Raphanus leaf tissue and a small amount of Phytophthora mycelium (around 1 mg of dry blotted mycelium) were co-chopped with a razor blade (Gilette) [30] in a Petri dish containing 500 µL extraction buffer. After chopping, the suspension was filtered through a 10 µm filter (CellTrics, Partec, Germany) and 2 mL of a Propidium Iodide staining solution was added. The samples were incubated overnight in the dark at 4°C. Measurements were done on a Partec PAS III flow cytometer (Partec, Germany) equipped with a 20 mW solid state laser (Sapphire 488-20) emitting at a fixed wavelength of 488 nm. The data were analyzed using Flomax software (Partec Münster, Germany). DNA content was calculated using the ratios between the peak positions of the Phytophthora sample and the Raphanus standard. The genome sizes are shown as pg DNA/2C in which 2C corresponds to the complete DNA content of the nucleus, irrespective of ploidy [31]. Histograms shown in this paper were made using Summit v4.3. The isolates were measured three times on different days over the course of three years. The coefficient of variation between the three repeated measures was on average 4.7%.
Oospore production and isolation. Oospore production was assessed for all hybrid isolates and for the parental species P. porri. Oospore production was induced by incubating a Phytophthora culture in a Petri dish containing V8 agar for one month at 15°C. Isolation of the oospores was done by digesting the mycelium using lysing enzymes from Trichoderma harzianum (Sigma). Ten mL of a solution containing 50 mg lysing enzymes was filter sterilized and preincubated overnight at 28°C. After this step, the solution was added to the Petri dish containing a Phytophthora culture and incubated for 2 days at 15°C. Hereafter, the solution containing oospores and mycelial fragments was scraped off the agar surface using a sterile spreader and filtered through a 70 µm filter (Cell Strainer, BD). Further removal of hyphal fragments was done by several rounds of centrifugation at low speed (1000 g), removing the supernatant and washing the pellet with sterile water. Oospores were visually examined under an Olympus BX51 microscope and the number of aborted oospores was counted.

Hybrid characterization
Polymorphic ITS sequences. The ITS region of 48 isolates (see Table 1) was amplified by primers ITS1 and ITS4 ( Table  2) and sequenced. The ITS sequence of 12 isolates showed double peaks in the DNA sequence chromatogram, which indicates that these isolates have a putative hybrid origin. Nine of these isolates originated from diseased Allium spp., mostly onion (Allium cepa). The other three isolates were obtained from diseased Parthenium argentatum, Chrysanthemum sp. and Pastinaca sativa. The isolates mainly originated from Japan (6 isolates) and Australia (3 isolates). The other three isolates were found in the USA, UK and Switzerland (Table 1).
For 11 of these isolates, the polymorphic base pairs were found exactly at those positions (7 on a total of 806 bp) where the ITS sequence of P. porri differs from that of P. primulae/P. taxon parsley (with the latter two having identical ITS sequences). These 11 hybrid isolates are shown in Figure 1, along with the possible parental species P. porri and P. primulae/P. taxon parsley, which showed no intraspecific or intra-isolate variation in ITS. Six isolates were polymorphic (indicated with a green colour in Figure 1) at all seven distinctive base pairs. The other five isolates showed polymorphisms in only some of the distinctive base pairs (nonpolymorphic sites are indicated with a blue or yellow colour in Figure 1). The non-polymorphic base pairs in a given isolate all corresponded to one of the two parental types, as if the ITS type is reverting back to one of the two parental states.
The 12 th putative hybrid, isolate CBS 112968, showed intraisolate polymorphisms at other positions in the ITS region and appeared not to be related to the other hybrid isolates. Instead, its ITS sequence was closely related to P. cichorii.
Recombinant ITS types in the hybrids point to genetic recombination between the different parental genomes. By cloning and sequencing the polymorphic ITS amplification products, it is possible to identify the distinct haplotypes coming from the different parents. We cloned the ITS PCR product of four putative hybrids and analyzed the sequence of 10 clones per isolate (see Figure 2a). Interestingly, we found that most of the clones showed recombinant sequences between the P. porri and P. primulae/P. taxon parsley parental haplotypes, suggesting that genetic recombination between the different parental genomes has occurred. ITS products for the possible parental species P. porri and P. taxon parsley were also cloned. Here, all clones had identical sequences, aside from an occasional random SNP, which was never at the distinctive positions.
To assess the possible occurrence of recombination caused by a PCR artifact, which has been reported in studies on plant hybrids [25], a PCR reaction was run on the combined genomic DNA of P. porri and P. taxon parsley. From this experiment, it is clear that recombinant ITS clones can be formed due to a PCR artifact (Figure 2a). However, the average number of crossovers per clone (shown in Figure 2 as 'average X per clone') was always higher in the hybrids compared to the combined genomic DNA experiment.
To eliminate the bias caused by this PCR artifact, we optimized the PCR protocol following the instructions suggested by Lahr and Katz [26]. This protocol was again tested on a mixture of DNA of the two parental states (P. porri + P. taxon parsley) and twenty clones were sequenced (see Figure 2b). Here, no recombinant ITS clones were detected. In the hybrid isolates, however, again different types of recombinant ITS clones were found, leaving no doubt that this is a real biological phenomenon.
Sequencing of Ypt1 haplotypes to determine the parental origin of the hybrids. The ITS region is located in the rRNA genes that are repeated in tandem arrays of several hundreds of copies. Because of this organization, the ITS region is sensitive to concerted evolution, i.e. the homogenization of the DNA sequence of the different repeats, which makes this region less suitable to study the parental origin of the interspecific hybrids. For this reason, we decided to sequence the highly variable Ras-related Ypt1 region [32], which is not sensitive to concerted evolution since it is a single copy gene.
We sequenced the Ypt1 region of 54 clade 8b isolates (see Table 1), including the hybrids. The resulting alignment length was 451 bp and a phylogenetic tree is shown in Figure 3. The interspecific variability in Ypt1 sequence was much higher compared to ITS. Despite the high interspecific variability, the Ypt1 locus showed almost no intraspecific variation and thus was very distinctive at the species level. The Ypt1 sequences of P. primulae and P. taxon parsley, however, were identical. Most hybrid isolates showed polymorphic Ypt1 sequences as observed by double patterns in the DNA sequence chromatogram. The only exception was isolate CBS 114040, which showed a sequence identical to that of P. primulae/P. taxon parsley. Remarkably, a polymorphic Ypt1 sequence was also observed in isolate ICMP14653. This isolate was also collected from diseased onion, but had an ITS sequence identical to that of P. primulae/P. taxon parsley, without polymorphisms. The Ypt1 gene was cloned for four hybrid isolates (CBS 138.87, ICMP14653, CBS 112968 and CBS 112969). For isolates ICMP14653, CBS 138.87 and CBS 112969, one haplotype was very similar to P. primulae/P. taxon parsley (0-2 SNPs) and the other haplotype was very similar to that of P. porri (2 SNPs). For isolate CBS 112968, two different haplotypes were recovered that were closely related but distinct from P. cichorii. The two distinct haplotypes for these isolates are shown in the phylogenetic tree in Figure 3. All other hybrid isolates showing a polymorphic Ypt1 sequence are not shown in Figure 3. mtDNA sequencing shows that at least three different hybridization events have occurred in clade 8b. Since mitochondria are inherited maternally through the oogonia [33], sequencing of the mtDNA can identify the species that acted as the maternal parent in the interspecific cross. We sequenced two genes (Cox1 and Nadh1) for 53 isolates and the appended alignment had a total length of 1539 bp. The multilocus phylogenetic tree is shown in Figure 4.
Phytophthora taxon parsley and most of the hybrid isolates (indicated as hybrid type 1 in Figure 4) form a cluster with high bootstrap support. Their closest relative was P. primulae which formed a separate cluster (99.7% sequence similarity, 5 SNPs). The hybrid isolate CBS 112969 (hybrid type 2 in Figure  4) had a different mtDNA sequence that showed most similarity     Figure 4), had a sequence most similar to that of P. cichorii (98.8 % sequence similarity, 18 SNP's), demonstrating that this hybrid was formed in a third hybridization event with an unknown species related to P. cichorii acting as the maternal parent. DNA content measurements suggest polyploidy in different Phytophthora clade 8b species and are consistent with (allo)polyploidy in the hybrids.
DNA content measurements were done for all isolates listed in Table  1 (except CBS 112969) and are presented in Table 3. Most P. porri isolates had a similar nuclear DNA content with a mean value of 0.270 pg DNA/2C (SE = 0.003, n = 14). One isolate (S05014(1)) contained approximately half of this DNA content (0.142 pg DNA/2C), suggesting polyploidy in this species, with the small DNA content most likely representing the diploid state. Moreover, four isolates contained a mixture of nuclei with two different DNA contents, with one nucleus type having a mean DNA content of 0.142 pg DNA/2C (SE = 0.002, n = 4) and the other having a mean DNA content of 0.273 pg DNA/2C (SE = 0.005, n = 4), suggesting that they are diploid/polyploid (probably diploid/tetraploid) heterokaryons. Histograms of the three different ploidy types are shown in Figure S1 (supporting data).
For P. primulae, all isolates measured had a similar DNA content with a mean value of 0.269 pg DNA/2C (SE = 0.005, n = 6). The two P. taxon parsley isolates both contained two types of nuclei and are assumed to be diploid/polyploid (probably diploid/tetraploid) heterokaryons. One isolate (BPIC2584) had a DNA content comparable to the heterokaryotic P. porri isolates described above. The other isolate (CBS 114156) had a considerably higher DNA content.
The 11 hybrid type 1 isolates had highly variable DNA contents ranging from 0.146 to 0.388 pg DNA/2C. There is a positive correlation between the relative amount of polymorphic positions in the ITS PCR amplicon (coloured green in Figure 1) and the DNA content for a given hybrid (type 1) isolate (Pearson's r, r = 0.690, n = 11, p = 0.019). This suggests that the hybrids evolve from a high DNA-content, allopolyploid state (such as in isolate CBS 126737) to a low DNA-content, diploid or diploid-like state (such as in isolate CBS 114040).
Most For comparison we also measured the DNA content of Phytophthora syringae. This species belongs to clade 8d [34] and is one of the closest relatives to the clade 8b species. Its mean DNA content was 0.176 pg DNA/2C (SE = 0.006, n = 3). High levels of aborted oospores in the hybrids point to meiotic problems. Oospore production was assessed in ten hybrid type 1 isolates as well as for hybrid type 2 and 3 and for three isolates of the parental species P. porri. In most hybrid type 1 isolates, oospore production was highly impaired. Three of the hybrid type 1 isolates did not produce oospores (CBS 138.87, CBS 112966 and CBS 114040); three isolates produced them in extremely low amounts (P6817, CBS 140.87 and CBS 126739). The other four isolates (ICMP 14653, CBS 114101, CBS 126738 and CBS 126737) produced oospores in higher amounts.
The number of abortive oospores was high in all six hybrid type 1 isolates that produced oospores (between 26.7% and 100%, see Table 4), and significantly higher than in isolates of Phytophthora porri (2.8% -8.6%). In Figure 5, examples of normal and aborted oospores are shown.
Hybrid type 2 (CBS 112969) produced oospores, but a high amount was aborted (78.6%). Hybrid type 3 also produced oospores, but in this case only a low amount of aborted oospores was detected (8.6 %).

Discussion
The increasing body of literature that is becoming available on hybridization in Phytophthora suggests that it could play an important role in the evolution of the genus. Natural interspecific hybridization events have been described in clade 1 [5][6][7]9,11], clade 6 [10], clade 7 [4,12] and now in clade 8 (this study). From these previous studies, it is clear that interspecific hybridization has the potential to create new strains that have a new or expanded host range. The most striking example remains Phytophthora alni, which caused a new disease on alder trees across Europe and quickly became one of the most serious threats in riparian ecosystems. The pathogen was proven to be of recent hybrid origin and contains an evolving group of allopolyploid genotypes [35].
The clade 8b hybrids were genetically characterized by cloning and sequencing two nuclear regions (ITS and Ypt1) and sequencing of two mtDNA genes (Cox1 and Nadh1). Hereby, three different types of hybrids could be distinguished with different parental species involved in their formation. The largest group (hybrid type 1, 11 isolates) originated from  hybridization between P. porri and P. taxon parsley. Hybrid type 2 (isolate CBS 112969) was closely related to the hybrid type 1, but has a different maternal parent which is an unknown species related to P. primulae. Hybrid type 3 (isolate CBS 112968) is unrelated to the first two types, and was formed in a hybridization event between two unknown species closely related to P. cichorii. There is a clear discrepancy between the nuclear (Ypt1) and mitochondrial (Cox1 + Nadh1) phylogenies. While most clade 8b species show a much faster mutation rate in the Ypt1 gene (as visualized by longer branch lengths) compared to the mtDNA genes, the reverse is true for P. porri and P. primulae. This demonstrates that gene flow (hybridization events) between P. porri and P. primulae is common in evolutionary terms. Moreover, hybrid types 1 and 2 both have mtDNA haplotypes different from P. primulae and P. porri, while their Ypt1 haplotypes are nearly identical to those of P. porri and P. primulae. This inconsistency is probably due to backcrossing or other reticulation events. We have measured DNA contents for over 50 isolates belonging to the different Phytophthora clade 8b species and for the hybrids. In P. porri, where 18 isolates were measured, we detected intraspecific DNA ploidy variation. The majority of the isolates were polyploid (probably tetraploid), one isolate was diploid, and four isolates were diploid/polyploid heterokaryons. Previous research in our lab [36] has shown that isolates belonging to these three different ploidy types, have identical AFLP patterns. In the case of allopolyploid hybrids, AFLP patterns would differ significantly from the parental patterns (i.e. a combination of AFLP fragments from both parents) [5]. Therefore, the data suggest that in P. porri, a whole genome duplication has occurred, creating an autopolyploid.
In P. primulae, no intraspecific DNA content variation was detected and the mean DNA content was similar to that of the polyploid P. porri isolates, suggesting that also P. primulae is polyploid. Both isolates of P. taxon parsley were diploid/ polyploid heterokaryons. In P. brassicae, three different DNA contents were found in the nine isolates that were analyzed. Most likely, these DNA content variations also represent ploidy variations. In the other clade 8b species, no considerable intraspecific DNA content variation was detected, but DNA contents were always in the polyploid range.
P. syringae (clade 8d), one of the closest relatives to the clade 8b species, had a DNA content of 0.176 pg DNA/2C. Isolates of P. ramorum (clade 8c) have also been evaluated for DNA content by flow cytometry using the same protocol as was used in our study [37]. These isolates had a stable DNA content of 0.146 pg DNA/2C (SE = 0.002, n = 3), which is assumed to represent the diploid state.
The hybrid type 1 isolates show a large variation in DNA content. This variation in DNA content was positively correlated with additivity (i.e. the presence of two alleles) in the nuclear loci. The isolates with the highest DNA content showed complete additivity in both nuclear loci, which is what is to be expected from a stable or recently formed allopolyploid. The isolates with the smallest DNA contents show almost no additivity in ITS, and in one isolate (CBS 114040) only one Ypt1 haplotype is still present. Processes that can cause this loss of additivity include chromosome loss, non-reciprocal translocations, gene conversion and unequal crossing over [18,38].
Recombinant ITS sequences between the two parental types were detected by cloning and sequencing of different hybrid isolates. Theoretically, these recombinations can be due to either asexual, mitotic recombination, or sexual recombination in the form of homeologous recombination. Asexual or mitotic recombination is in general a rare event that is associated with double strand break repair [39]. Since the hybrid isolates also exhibit problems in meiosis, as shown by high numbers of aborted oospores, homeologous recombination is likely to have occurred. Although it is unclear to what extent homeologous recombination occurs in other parts of the genome, it can be an important process of creating novel genetic variation as has been described for plant hybrids [18].
Our study raised some concern on the use of ITS as the universal barcode for identification of Phytophthora, since hybridization events can be easily overlooked. The ITS region is located in the rRNA genes that are repeated in tandem arrays of several hundreds of copies. Because of this organization, the ITS region is sensitive to concerted evolution, i.e. the homogenization of the DNA sequence of the different repeats. From plant research, it is known that the ITS can evolve in three different ways after interspecific hybridization: (1) unidirectional concerted evolution leads to loss of one of the parental copies and fixation of the other (2) formation of a new ITS type that represents a mixture of the two parental types, (3) presence of both ITS types, which is the case in young hybrid taxa, although this situation can also be stable in older hybrid taxa [40][41][42]. In the clade 8b hybrids, unidirectional concerted evolution is ongoing, keeping one of the two parental haplotypes and in this way erasing the footprints of hybridization.
Most Phytophthora species show a certain degree of host specificity, and this is also the case for the clade 8b species. Hence, the question remains how and where different species can meet intimately, a requirement for interspecific hybridization to occur. Since the clade 8b species show a clear host preference for winter grown field vegetables that are commonly used in crop rotation systems (such as leek, carrot, cabbages, lettuce, parsley, chicory), it is likely that the different species can be present on the same field. Moreover, all three hybrid types, formed in hybridization events with different parental species, have been isolated from onion (Allium cepa). This led us to hypothesize that Allium cepa might be a common susceptible host of the different clade 8b species involved in the hybrids' formation, offering a hot spot for interspecific hybridization to occur. Pathogenicity tests with P. porri, one of the two parents of hybrid type 1 and 2, show that it can easily infect onion (unpublished results). Pathogenicity tests with the other clade 8b species on onion, could shed more light on this hypothesis.
Of the hybrid type 1, six isolates were derived from Allium cepa, the other isolates were derived from Allium victorialis, Allium grayi, Pastinaca sativa, Chrysanthemum sp. and Parthenium argentatum, indicating a clear expansion of host range compared to the parents that were so far only detected on leek (P. porri) and parsley (P. taxon parsley).
As a result of globalization, plants and therefore their pathogens are now being traded all over the world on a large scale. This evidently increases the chance of interspecific hybridization since closely related species or different genotypes of the same species that have been geographically isolated are brought together while their reproductive barriers might be incomplete [43].
Therefore, the implications of hybridization on the adaptive potential and pathogenicity of Phytophthora spp. deserve to be studied in more detail. An outstanding topic of discussion in this respect is the fate of genes involved in plant pathogen interaction. In the past few decades, it has become clear that Phytophthora pathogens are extremely flexible and this has made it a very challenging task for plant breeders to develop crops with durable resistance against Phytophthora. A lot of attention has been given to the biology of the RXLR effectors that have been discovered recently [44,45]. These effector genes have been coined as one of the major determinants of host specificity. From whole genome sequencing, we know that a single pathogenic Phytophthora isolate can contain high numbers of different and often unique RXLR effector genes (563 different RXLR effector genes predicted from the P. infestans genome sequence [45]). It has also been discovered that the genome of P. infestans (with its 250 Mb the largest oomycete and chromalveolate genome sequenced up till now) consists for 74% of repetitive sequence, contributing to the flexibility of the genome.
It is easy to imagine how (allopolyploid) hybridization and its consequences could have helped to build up the enormous effector repertoires that we find in Phytophthora pathogens today. The fact that effector genes are mainly contained in gene sparse regions in the DNA enables recombination between effector gene regions by homeologous recombination, without high risks of lethality.
Intra-specific hybridization between divergent races of the same species could also enhance the pathogen's pathogenic potential by expansion and/or recombination of effector gene repertoires. In P. ramorum, a heterothallic species, crosses have been induced between different genotypes that have diverged as a consequence of geographical isolation [46]. The resulting oospore progenies were characterized by DNA content estimation and microsatellite genotyping [37]. A large proportion of the progeny showed DNA contents that were considerably larger (up to 70%) than their parental DNA content. Microsatellite patterns also indicated polyploidy and aneuploidy in the progeny.
The same processes might be responsible for the increased pathogenic potential observed for P. infestans in Europe since the 1980s, when sexual reproduction of the pathogen was enabled by the introduction of the A2 mating type from its center of origin to Europe. In a recent study [47], herbarium samples of the P. infestans strains that caused the Irish potato famine in the 19 th century were collected and their genomes were sequenced. These genomes were compared to the genomes of recent aggressive P. infestans strains that have outcompeted the old strains. An important difference between the old strains (that were dominant for about 50 years) and the new strains, is the rise in ploidy level. The old strains were found to be diploid, while the new strains are mostly triploid and sometimes tetraploid, which was estimated based on allele frequency from the resequencing alignments. It is unclear what caused this rise in ploidy level, but in our opinion, hybridization between divergent P. infestans genotypes, seems to be a likely scenario.
In conclusion, we have found that interspecific hybridization has occurred different times in clade 8b with different parental species involved. The combination of sequencing different genes and estimating DNA content by flow cytometry provided valuable insights in the processes involved in and after interspecific hybridization. A recently formed hybrid will contain the complete complement of both parental genomes (allopolyploidy). Novel gene combinations such as a combined effector repertoire can enable the pathogen to invade new habitats or hosts. After allopolyploidization, different genetic processes such as chromosome loss and chromosomal rearrangements (which can be between the different parental genomes through homeologous recombination), can enable the pathogen to further adapt to a new habitat or host. As can be seen in the hybrid type 1 isolates, the progeny of a single hybridization event can follow different evolutionary paths, which ultimately can result in the formation of new species. It is clear from our data that the footprints of hybridization can easily be lost. Therefore, it is possible that the occurrence and importance of hybridization in Phytophthora is highly underestimated. We hypothesize that the occurrence of DNA content variation and polyploidy in Phytophthora species is linked with (ancient) hybridization events. Figure S1. Ploidy variation in P. porri. DNA content measurements on mycelial nuclei of three different P. porri isolates using the internal standard Raphanus sativa cv. Saxa (2C = 1.11 pg). A, logarithmic histogram with 'diploid' P. porri (isolate S05014(1)); B, histogram with 'tetraploid' P. porri (isolate CBS 127098); C, histogram with heterokaryotic 'diploid/ tetraploid' P. porri (isolate S05007(2)); Pp1: P. porri G1 peak; Pp2, P. porri G2 peak (panels A and B) or second G1 peak (panel C); Pp3: P. porri G2 peak; Rs1: Raphanus sativa G1 peak; Rs2: Raphanus sativa G2 peak. (PPTX)